Table of Contents
Mice were bred and housed in specific pathogen-free facilities at The Francis Crick Institute. All protocols for breeding and experiments were approved by the Home Office (UK) under project license P4D8F6075 and performed in accordance with the Animal Scientific Procedures Act, 1986.
All DNA constructs were produced using Escherichia coli DH5a (Thermo Fisher Scientific) and extracted using a plasmid midiprep kit from Qiagen. The plasmids used in this study were: RFP-GAL-350, LAMP1-RFP (Addgene, 1817) and G3BP1-mEGFP (Addgene, 135997).
iPS cell and iPSDM culture
KOLF2 human induced pluripotent stem (iPS cells were sourced from Public Health England Culture Collections (catalogue number 77650100). mEGFP-G3BP1 human iPS cells (used for CLEM studies) were sourced from Coriell Institute (catalogue number AICS-0082 cl.1). iPS cells were maintained in Vitronectin XF (StemCell Technologies) coated plates with E8 medium (Thermo Fisher Scientific). Cells were authenticated by STR profiling upon receipt and are checked monthly for Mycoplasma contamination by PCR. Cells were passaged 1:6 once at 70% confluency using Versene (Gibco). Monocyte factories were set up following a previously reported protocol42. In brief, a single-cell suspension of iPS cells was produced with TryplE (Gibco) at 37 °C for 5 min and resuspended in E8 plus 10 µM Y-27632 (Stem Cell Technologies) and seeded into AggreWell 800 plates (StemCell Technologies) with 4 × 106 cells per well and centrifuged at 100g for 3 min. The forming embryonic bodies were fed daily with two 50% medium changes with E8 supplemented with 50 ng ml−1 human BMP4 (Peprotech), 50 ng ml−1 human VEGF (Peprotech) and 20 ng ml−1 human SCF (Peprotech) for 3 days. On day 4, the embryonic bodies were collected by flushing out of the well with gentle pipetting and filtered through an inverted 40-µm cell strainer. Embryonic bodies were seeded at 100–150 embryonic bodies per T175 or 250–300 per T225 flask in factory medium consisting of X-VIVO15 (Lonza) supplemented with Glutamax (Gibco), 50 µM β-mercaptoethanol (Gibco), 100 ng ml−1 hM-CSF (Peprotech) and 25 ng ml−1 human IL-3 (Peprotech). These monocyte factories were fed weekly with factory medium for 5 weeks until plentiful monocytes were observed in the supernatant. Up to 50% of the supernatant was collected weekly and factories fed with 10–20 ml factory medium. The supernatant was centrifuged at 300g for 5 min and cells resuspended in X-VIVO15 supplemented with Glutamax and 100 ng ml−1 human M-CSF and plated at 4 × 106 cells per 10-cm petri dish to differentiate over 7 days. On day 4, a 50% medium change was performed. To detach cells, iPSDM plates were washed once with PBS then incubated with Versene for 15 min at 37 °C and 5% CO2 before diluting 1:3 with PBS and gently scraping. Macrophages were centrifuged at 300g and plated for experiments in X-VIVO15.
U2OS and HeLa cells
U2OS and HeLa cells were cultured using standard conditions. Both cell lines were maintained in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS). The cells were incubated at 37 °C in a humidified atmosphere containing 5% CO2. U2OS wild-type and G3BP1/2 KO U2OS cell lines were from the laboratory of P. Anderson37. Cells were negative for mycoplasma contamination (checked by PCR).
Plasmid DNA was electroporated into iPSDM using the Neon system (Invitrogen). iPSDM were resuspended at 1.5 × 106 cells in 100 µl buffer R. Ten microlitres of cells per 1 µg plasmid DNA mix was aspirated into a Neon pipette and electroporated in electroporation buffer E at 1,500 V for 30 ms with 1 pulse. Cells were then plated in ViewPlate glass bottom 96 well plates (6005430, PerkinElmer) for high-content analysis or in IBIDI µ-Slide 18-well glass bottom coverslips (81817) for confocal imaging studies.
iPSDM were washed twice with PBS and electroporated in the appropriate primary nucleofection solution (AmaxaTM Human Monocyte Nucleofector Kit, VPA-1007) using the Lonza 2b Nucleofector (Nucleofector 2b Device, AAB-1001). Five million iPSDMs were used per reaction and resuspended in 100 µl of primary nucleofection solution containing 4 µg of Cas9 from Streptococcus pyogenes (SpCas9) (IDT) mixed with a total of 12 µg of targeting synthetic chemically modified single guide RNA (sgRNA) (Synthego) (outlined below). iPSDMs were then nucleofected with two sgRNAs targeting the G3BP1 genes and two sgRNAs targeting the gene G3BP2, and the Cas9–RNP mix using the Y001 program. Nucleofected cells were cultured in pre-warmed X-VIVO15 in a 35 mm Sterilin petri dish plate (121 V, Thermo Scientific). At 2 h post nucleofection, 100 ng ml−1 human M-CSF was added to the cells. Dishes were incubated in a humidified 37 °C incubator with 5% CO2. After 3 days, an equal volume of fresh complete media including 100 ng ml−1 human M-CSF was added. Six days after the initial isolation, differentiated macrophages were detached in Versene and plated for experiments in X-VIVO15.
sgRNAs used were: G3BP2_g1: CGCCCTACAAGCAGCGGACT; G3BP2_g2: AAGCTCCGGAATATTTACAC; G3BP1_g1: CGCCCGACCAGCAGGGGACT; and G3BP1_g2: AGGCCCCAGACATGCTGCAT.
siRNA transfection of HeLa cells
HeLa cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% FBS. For short interfering RNA (siRNA) transfection, SMARTpool siRNA targeting G3BP1 and G3BP2 (obtained from Dharmacon) were diluted in Opti-MEM medium, and DharmaFECT transfection reagent was mixed with the siRNA to achieve a final concentration of 25 nM. The siRNA transfection complex was incubated for 5–10 min at room temperature. HeLa cells were plated at a density allowing for 50–70% confluency at the time of transfection. The siRNA transfection complex was then added to the cells plated in a 6 well plate (1.5 ml final volume) and incubated for 24 h before repeating the same protocol for an extra day (two rounds of transfection). After 24 h, cells were collected for subsequent experiments. As a control siRNA, a Silencer Select Negative Control No. 2 siRNA (4390846, Thermo Fischer) was used following the same protocol.
The targeting siRNAs were as follows:
ON-TARGETplus SMARTpool siRNA J-012099-06, G3BP1 target sequence GUGGUGGAGUUGCGCAUUA; ON-TARGETplus SMARTpool siRNA J-012099-07, G3BP1 target sequence AGACAUAGCUCAGACAGUA; ON-TARGETplus SMARTpool siRNA J-012099-08, G3BP1 target sequence GAAGGCGACCGACGAGAUA; ON-TARGETplus SMARTpool siRNA J-012099-09, G3BP1 target sequence GCGAGAACAACGAAUAAAU; ON-TARGETplus SMARTpool siRNA J-015329-09, G3BP2 target sequence UGAAUAAAGCUCCGGAAUA; ON-TARGETplus SMARTpool siRNA J-015329-10, G3BP2 target sequence GAAUUUAAGUCUGGGACGA; ON-TARGETplus SMARTpool siRNA J-015329-11, G3BP2 target sequence ACAACGACCUAGAGAACGA; ON-TARGETplus SMARTpool siRNA J-015329-12, G3BP2 target sequence: GCGAUGGUCUUGACUAUUA.
Human monocyte-derived primary macrophage isolation
White blood cells were isolated from leukocyte cones (NC24) supplied by the UK National Health Service blood and transplant service by centrifugation on Ficoll-Paque Premium (GE Healthcare 17-5442-03) for 60 min at 300g. Mononuclear cells were washed twice with MACS rinsing solution (Miltenyi 130-091- 222) to remove platelets, then remaining red blood cells were lysed by incubation at room temperature with 10 mL RBC lysing buffer (Sigma R7757) per pellet for 10 min. Cells were washed with rinsing buffer and pelleted once more, then resuspended in 80 µl MACS rinsing solution supplemented with 1% BSA (MACS/BSA) and 20 µl anti-CD14 magnetic beads (Miltenyi 130-050-201) per 108 cells and incubated on ice for 20 min. Cells were then washed in MACS/BSA by centrifugation, resuspended in 500 µl MACS/BSA per 108 cells and passed through an LS column (Miltenyi 130-042-401) in the field of a QuadroMACS separator magnet (Miltenyi 130-090-976). The column was washed three times with MACS/BSA, then positively selected cells were eluted, centrifuged and resuspended in RPMI 1640 with GlutaMAX and HEPES (Gibco 72400- 02), 10% heat-inactivated FBS and 50 ng ml−1 hM-CSF (Preprotech) to a concentration of 106 cells per ml in untreated petri dishes. These were placed in a humidified 37 °C incubator with 5% CO2, with an equal volume of fresh media including human M-CSF added after 3 days. Six days after the initial isolation, differentiated macrophages were detached in 0.5 mM EDTA in ice-cold PBS and 5 × 105 cells seeded per well of a 12-well plate for western blot experiments.
Mtb infection of macrophages
Wild-type Mtb H37Rv and Mtb H37Rv ΔRD1 were provided by D. Young and S. H. Wilson. Fluorescent Mtb strains were generated as previously reported51. E2-Crimson Mtb was generated by transformation with pTEC19 (Addgene 30178, deposited by L. Ramakrishnan). Strains were verified by sequencing and tested for phthiocerol dimycocerosate positivity by thin layer chromatography of lipid extracts from Mtb cultures. Mtb strains were cultured in Middlebrook 7H9 supplemented with 0.2% glycerol, 0.05% Tween-80 and 10% albumin dextrose catalase (ADC). For macrophage infections, Mtb was grown to an optical density (OD600) ∼ 0.8 then centrifuged at 2,000g for 5 min. The pellet was washed twice with PBS, then the pellet was shaken with 2.5–3.5 mm glass beads for 1 min to produce a single-cell suspension. The bacteria were resuspended in 10 ml cell culture medium and centrifuged at 300g for 5 min to remove clumps. The OD600 was determined, and bacteria diluted to an appropriate OD for the required MOI—assuming OD600 = 1 equates to 108 bacteria per ml—before adding to cells in a minimal volume. After 2 h, the inoculum was aspirated, cells washed twice with PBS and fresh culture medium added. Cells were then incubated for appropriate time points before collecting for analysis as described in the sections below.
A 333 mM stock of LLOMe (Cat# 4000725, Bachem) was prepared in methanol and frozen at −20 °C in tightly sealed tubes. For LLOMe treatment, the medium was replaced with X-VIVO15 (iPSDM), RPMI (HMDM) or DMEM (HeLa and U2OS cells) containing 1 mM of LLOMe unless otherwise indicated in the figure legend. Methanol (0.3%) in the respective cell culture media was used in all control conditions.
A solution of crystalline silica (MIN-U-SIL-15, US Silica) at 200 μg ml−1 was prepared in X-VIVO15 and the cells were stimulated for 3 h, after which they were processed for downstream applications. In experiments with BAFA1 (Merck, B1793-10UG), iPSDM were pre-incubated for 1 h with a 100 nM solution. BAFA1 was kept in the medium during silica treatment.
For lysis, cells were washed once with PBS, and lysed on ice in RIPA buffer (Millipore) containing complete, EDTA-free protease inhibitor (Roche). The samples were boiled at 95–100 °C for 5 min in LDS sample buffer and reducing agent (NuPAGE, Life Technologies) and run on a NuPAGE 4–12% Bis-Tris gel (Life Technologies). The gels were transferred onto a PVDF membrane using an iBlot 2 Dry Blotting System (Thermo Fischer), program P0. Membranes were blocked in 5% skimmed milk powder in PBS plus 0.05% Tween-20 (PBS-T) for 1 h at room temperature then incubated with primary antibody overnight at 4 °C. Membranes were washed in PBS-T and incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Membranes were developed with enhanced chemiluminescence reagent (BioRad) and imaged on an Amersham Imager 680 instrument. Antibodies used: anti-G3BP1 (13057-2-AP) and anti-G3BP2 (16276-1-AP) from Proteintech, anti-p-eIF2α (Ser51) (9721), anti- eIF2α (9722) and anti-β-Actin (8H10D10, 12262) from Cell Signalling Technology; and HRP-conjugated anti-mouse (W4021) and anti-rabbit (W4011) antibodies from Promega. All antibodies were used at 1:1,000 dilution with the exception of HRP-conjugated antibodies that were used at 1:10,000 dilution.
In vitro condensate–membrane experiments
1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) and 1,2-dioleoyl-sn-glycero-3-phospho-l-serine (DOPS) were purchased from Avanti Polar Lipids. The fluorescent lipid dye ATTO 647N-DOPE was acquired from ATTO-TEC. Polyvinyl alcohol (PVA, with MW 145000) and chloroform (HPLC grade, 99.8%) were purchased from Merck. Lipid stocks were mixed as chloroform solutions at 4 mM, containing 0.1 mol% dye, and were stored until use at −20 °C. Fluorescein isothiocyanate isomer (FITC), poly (A) RNA, 4-(2-hydroxyethyl)−1-piperazineethanesulfonic acid buffer (HEPES), sucrose, glucose, carbonate buffer tablets, hydrochloric acid (HCl), sodium hydroxide (NaOH), magnesium chloride (MgCl2), dimethyl sulfoxide (DMSO) and sodium chloride (NaCl) were obtained from Sigma-Aldrich. All solutions were prepared using ultrapure water from a SG water purification system (Ultrapure Integra UV plus, SG Wasseraufbereitung) with a resistivity of 18.2 MΩ cm.
Preparation of glycinin was achieved as described52. In brief, defatted soy flour was dispersed in 15-fold water (weight) and pH was adjusted to 7.5 using a 2 M NaOH solution. Afterwards, the dispersion was centrifuged for 30 min at 9,000g at 4 °C. Dry sodium bisulfite (SBS) was added to the supernatant (0.98 g l−1 SBS), and the pH of the solution was adjusted to 6.4 with a solution of 2 M HCl and kept at 4 °C overnight. Then the dispersion was centrifuged during 30 min at 6,500g at 4 °C. The obtained glycinin-rich precipitate was dispersed in fivefold water and the pH was adjusted to 7.4. The glycinin solution was then dialysed against Millipore water for two days at 4 °C and then freeze-dried to acquire the final product with a purity of 97.5%. For glycinin labelling, a 20 mg ml−1 of protein solution was prepared in 0.1 M carbonate buffer (pH 9). A FITC solution (4 mg ml−1 in DMSO) was slowly added into the protein solution with gentle stirring to a final concentration of 0.2 mg ml−1. The sample was incubated in the dark while stirring at 23 °C for 3 h. The excess dye was removed using a PD-10 Sephadex G-25 desalting column (GE Healthcare), and buffer was exchanged for ultrapure water. The pH of the labelled protein solution was adjusted to 7.4 by adding 0.1 M NaOH. For fluorescence microscopy experiments, an aliquot of this solution was added to the working glycinin solution to a final concentration of 4%.
G3BP1 expression and purification were adapted from Guillén-Boixet et al.10 with modifications. In brief, recombinant His6–G3BP1–MBP and His6–GFP–G3BP1–MBP were expressed in and purified from SF9 insect cells (Expression Systems, 94-001 F) using a baculovirus expression system53. Following cell lysis (LM10, Microfluidics) in buffer A (50 mM HEPES/NaOH (pH 7.5), 1 M NaCl, 5% (w/v) glycerol, 2 mM EDTA, 2 mM DTT) supplemented with 1× EDTA-free protease inhibitor cocktail (Roche), the lysate was cleared by centrifugation at 50,000g for 50 min and subsequently the supernatant was applied to a custom casted HiFlow amylose column (New England Biolabs). The column was washed with 10 column volumes buffer A, followed by 10 column volumes buffer B (50 mM HEPES/NaOH, 100 mM NaCl, 5% gylcerol (w/v), 2 mM EDTA, 2 mM DTT, pH 7.5). 150 μl of 1 mg ml−1 His-tagged 3 C protease (home-made) in buffer B were applied for on-column tag removal and incubated for 8 h at room temperature. The tag-free product was washed out with buffer B, applied to a 5 ml HiTRAP Q HP column (Cytiva) equilibrated in buffer B and eluted with a linear gradient against buffer A. Fractions containing the target protein were applied to a HiLoad Superdex 200-pg 26/600 column (Cytiva) equilibrated in buffer C (50 mM HEPES/NaOH, 150 mM NaCl, 5% gylcerol (w/v), 2 mM EDTA, 2 mM DTT pH 7.5). G3BP1 fractions were pooled, concentrated, flash frozen with liquid nitrogen and stored at −80 °C.
DOPC and DOPC:DOPS 9:1 GUVs containing 0.1 mol % ATTO 647N-DOPE were prepared by the PVA gel-assisted method54 which allows protein encapsulation during vesicle swelling. In brief, 2 coverslips were cleaned with water and ethanol and dried with nitrogen. PVA solution was prepared by diluting PVA in deionized water at 60 °C to a final concentration of 40 mg ml−1. A small aliquot (20–50 µl) of the PVA solution was spread on the glass slides and dried for 1 h at 60 °C. Lipid stock solution (3–4 μl) were spread on the PVA-coated glasses and kept for 1 h under vacuum at room temperature. The chamber was assembled using the two coverslips and a 2-mm-thick Teflon spacer, and filled as follows according to the different experiments:
For G3BP1 encapsulation in GUVs, the swelling buffer consisted in 10 µM GFP–G3BP1 in 50 mM HEPES pH 7.5, 100 mM NaCl, 4 mM MgCl2.
For glycinin encapsulation, the swelling solution contained 10 mg ml−1 glycinin in 15 mM NaCl, 10 mM sucrose at pH 7.4 for the pH-triggered phase separation experiments. For inducing salt-driven phase separation the solution consisted of 300 mM sucrose at pH 7.4.
In all cases pH was adjusted using a 2 M NaOH solution. After 30 min, the chamber was opened, and the vesicles were collected carefully to prevent PVA detaching from the cover glasses. Osmolarities were measured and adjusted using a freezing point osmometer (Gonotec Osmomat 3000).
Encapsulation of the proteins was preferred over having the proteins in the external solution due to several experimental constraints. First, protein concentration is a limiting factor and having the protein inside the vesicles requires much less protein than having it in the external solution, which in the microfluidic experiments is under a constant flow. Second, if the proteins were in the external solution, the labelled proteins would make the complete field of view fluorescent, which would hinder visualization of the experiment and high-precision imaging. Finally, it is preferable to have the lower pH on the outside, because GUV formation leads to higher yield and better quality at neutral rather than acidic pH. It is important to note that the results are independent of the experimental choice for solution location (inside versus outside), since condensate formation is driven by the external and internal solutions mixing at the pore (as shown in Fig. 2 and Extended Data Fig. 4), and the outcome will not depend on the positional order of the solutions across the membrane. Furthermore, because of the large size of the GUVs, the membrane is almost flat with negligible curvature (lower than 0.1 μm−1) and not expected to have a role.
Microfluidics and solution exchange
The microfluidic device consists of a cascade GUV trapping system, which is described in detail elsewhere55. It was produced using poly(dimethylsiloxane) (PDMS) pre-cursor and curing agent (Sylgard 184, Dow Corning GmbH), at a mass ratio of 10:1. After polymerization at 80 °C for 3 h, inlet and outlet holes were punched with a biopsy punch with a plunger system (Kai Medical). Then the PDMS chips and glass coverslips were treated with plasma for 1 min using high-power expanded plasma cleaner (Harrick Plasma), and then bonded together. Before the experiments, the desired amount of solution was filled into the microfluidic device by centrifugation at 900 relative centrifugal force (Rotina 420 R, Hettich). Solution exchange was performed with a NeMESYS high-precision syringe pump. The flow speed for the initial washout of the remaining labelled protein in the exterior (Extended Data Fig. 4f,g) was set to 1 μl min−1 for 40 min to ensure at least 10 times exchange of the internal volume of the microfluidic device (~4 µl). To change the pH and introduce the hypotonic solutions, a higher speed of 2 μl min−1 was used to ensure rapid exchange, to avoid phase separation due to passive proton permeation56,57,58. At this high flow, vesicles were pushed against the posts of the microfluidic device and small deformations could be observed (Fig. 2a). After complete solution exchange, the flow speed was lowered to 0.035 μl min−1 to prevent vesicle movement and facilitate imaging.
The hypotonic buffer consisted in 20 mM HEPES, 100 mM NaCl, 4 mM MgCl2, pH 5 for G3BP1 experiments. For glycinin pH-driven phase separation the external hypotonic solution was deionized water at pH 4.8, with the pH adjusted using a solution of 2 M HCl. For the experiments with phase separation triggered by NaCl, the hypotonic buffer consisted of a 100 mM NaCl solution, pH 7.4.
In vitro G3BP1–RNA granules formation
G3BP1–poly(A)-RNA condensates were reconstructed following Guillén-Boixet et al.10. To form the condensates, 20 µl of a solution containing a final concentration of 5 µM GFP–G3BP1 and 200 ng µl−1 of poly(A)-RNA in 50 mM HEPES, 100 mM NaCl, 4 mM MgCl2 was directly placed on a glass slide and immediately visualized by confocal microscopy.
To simulate poration and plugging of lipid vesicles, we consider a minimal coarse-grained molecular dynamics system where the solution of proteins in and out of the vesicle is described by beads dispersed in implicit solvent. The membrane is described as a one-bead-thick fluid elastic layer of mechanical properties that mimic biological membranes. We tune the interactions between solution beads in the different compartments to capture the differences between the cytoplasmic and endosomal protein solutions and explore how the system responds to different degrees of membrane poration. Details on how the molecular dynamics was performed are in the Supplementary Information and Supplementary Table 1.
Molecular dynamics analysis
To characterize and understand the plugging effect of protein condensates on damaged vesicles we can track how the vesicle surface evolves and how the particles which are initially in each compartment (inside or outside) mix throughout the course of the simulation.
To do so, a first essential step is reconstructing the membrane surface from the model particle positions and orientations, for which we use custom Python code59 based on the ball pivot algorithm60. Our analysis provides an estimate for the vesicle surface area and local boundaries and orientations together with a collection of pores and their perimeter, which allows us to evaluate pore closure.
Once we have defined the vesicle surface, we evaluate condensation by running a clustering algorithm on the solute and protein particles (part of the Ovito library in Python61). This allows us to distinguish between condensate particles and solution particles and to calculate droplet properties (such as mass or radius of gyration) as well as the total droplet growth rate over time. Once we have identified the particles that remain in solution, we can then determine which compartment they belong to (inside the vesicle or outside) and by comparing with the initial conditions we can determine solution exchange fluxes.
To classify the particles as inner lumen or outer lumen we determine each particle’s closest six surface triangles (using a neighbour list algorithm from the Freud library for Python62) and compute the unit vector along the line connecting the particle and each of its neighbour triangles centres. We then compute the dot products ρ of these unit vectors with the local normal of the surface (previously identified using the ball pivot algorithm). Finally, if the average dot product ⟨ρ⟩ > 0.5 we classify the particle as being inside the vesicle. Conversely, if ⟨ρ⟩ < −0.5 we classify the particle as being outside the vesicle. In this analysis, particles that are close to the pore present dot products close to zero (|⟨ρ⟩| < 0.5), and in this case we compare their position to the pore’s centre of mass to determine their compartment.
Once we have a list of inner particles, outer particles and droplet particles, we can define fluxes by comparing these values across the simulation and characterize the mixing of solutions and the condensate growth in different conditions.
Stress granules and GAL-3 staining
After the indicated treatment, cells were washed once with PBS and fixed with 4% methanol-free paraformaldehyde (PFA) in PBS for 15 min. After 3 washes with PBS, cells were permeabilized using a 0.5% Triton X-100 (Sigma)/PBS solution for 10 min. Cells were then immunostained using the corresponding antibodies. Antibodies were: anti-G3BP1 (13057-2-AP) or alternatively, anti-G3BP1 (66486-1-Ig, sc-365338 AF546), anti-G3BP2 (16276-1-AP) anti-TIA1 (12133-2-AP), anti-PABPC1 (10970-1-AP), anti-ALIX (12422-1-AP), anti-CHMP2a (10477-1-AP), anti-CHMP4b (13683-1-AP), anti-annexin A1 (66344-1-Ig), anti-annexin A2 (66035-1-Ig), anti-EIF3B (10319-1-AP), and anti-EIF4G1 (15704-1-AP) from Proteintech. Anti-GAL-3 (125410) and anti-LAMP-1 (121610) from Biolegend. Anti-p62 (GTX111393) from GeneTex, anti-phospho TBK1 (5483 T) from Cell Signalling Technologies. Anti-PI4K2a (B-5, sc390026) and anti-ORP9 (A-7, sc398961) from Santa Cruz. Antibodies were used at 1:200 dilution. Images were acquired on Ibidi glass bottom slides (81817) using a Leica SP8 confocal microscope (for Mtb fixed imaging and silica crystals experiments) or using a VT-iSIM superresolution imaging system (Visitech International). Nuclear staining was performed using 300 nM DAPI (Life Technologies, D3571) in PBS for 5 min. The area of GAL-3 and stress puncta per cell was evaluated using the image analysis software FIJI/ImageJ as described below.
Fluorescence recovery after photobleaching
Fluorescence recovery after photobleaching (FRAP) experiments were performed using a Zeiss Invert880 microscope system (Zeiss) and the bleaching module of Zeiss ZEN software. Photobleaching of G3BP1–mEGFP was achieved using a 488-nm laser at a scan speed = 1 (pixel dwell 131.07 ms). Images were captured at 20 or 30-s intervals for at least 20 min. The photobleaching started after three time points that were used as the basal intensity reference.
Confocal microscopy of GUVs
Confocal Leica SP8 microscope equipped with a 63×, 1.2 NA water-immersion objective and a 40× 1.3 NA oil immersion objective (Leica) was used for imaging. FITC and ATTO 647N-DOPE were excited using the 488 nm and 633 nm laser lines, respectively and signal was collected between 500–600 nm for FITC and 650–720 nm for ATTO. Time-lapse sequences (xyt) of individual GUVs were acquired at 650 ms per frame using the bidirectional acquisition mode.
High-content live-cell imaging
Thirty thousand iPSDM were seeded into a ViewPlate glass bottom 96 well plate (PerkinElmer) and treated with LLOMe or infected with Mtb as described above. The plate was sealed with parafilm and placed in a pre-heated (37 °C) Opera Phenix microscope with a 40× or 60× water-immersion lens (PerkinElmer) with 5% CO2. Capture settings were: LysoTracker Red was excited with the 561 nm laser at 10% power with 100 ms exposure. MitoTracker Deep Red FM (M22426, Thermo Fischer), iABP probe and Mtb E2-Crimson were excited with the 640 nm laser at 10% power with 100 ms exposure. DAPI was excited with the 405 nm laser at 20% power with 100 ms exposure. At least 20 fields per well were imaged in all the experiments. Images were acquired at 1,020 × 1,020 pixels using Harmony 4.9 high-content imaging and analysis software (PerkinElmer).
Superresolution live-cell imaging
iPSDMs transfected with GAL-3–RFP, G3BP1–mEGFP or LAMP1–RFP were treated with LLOMe (1 mM) and imaged on a VT-iSIM superresolution imaging system (Visitech International), using an Olympus IX83 microscope, 150×/1.45 Apochromat objective (UAPON150XOTIRF), ASI motorized stage with piezo Z, and 2× Prime BSI Express scientific CMOS cameras (Teledyne Photometrics). Cells were always in the stage incubator at 37 °C and 5% CO2. Simultaneous GFP and mCherry imaging was done using 488 nm and 560 nm laser excitation and ET525/50 m and ET600/50 m emission filters (Chroma), respectively. Z-stacks (100 nm z-step) were acquired at the intervals indicated in the figure legends. The microscope was controlled with CellSens software (Olympus). Image processing and deconvolution was done using Huygens Essential software (Scientific Volume Imaging). For 3D imaging, spatial deconvolution and 3D surface-rendering reconstruction, z-stack slices were defined each 200 nm (if confocal microscope was used) or 100 nm (if VT-iSIM microscope was used) and images were processed using Huygens Essential Software (Scientific Volume Imaging).
LysoTracker recovery assay
Forty thousand cells (background indicated in the figure legends) were seeded into a ViewPlate glass bottom 96 well plate (PerkinElmer). Cells were loaded with the nuclear dye Nuclear Green LCS1 (Abcam) at a dilution of 1:5,000 and 25 nM LysoTracker DND-99 (Thermo Fisher Scientific) for 20 min. The cells were washed twice in X-VIVO15, and the medium was replaced with X-VIVO15 containing 25 nM LysoTracker. The cells were imaged every 1 min at 37 °C, 5% CO2 using an Opera Phenix microscope (PerkinElmer). First, a baseline was established by imaging 3 time points, followed by the addition of LLOMe to a final concentration of 1 mM. After 2 time points, the cells were washed 3 times with X-VIVO15 and the medium was replaced with X-VIVO15 containing 25 nM LysoTracker, and lysosomal recovery was followed for 20 min. Analysis was done as indicated below (see ‘Imaging analysis’).
10K-dextran release assay
Wild-type and G3BP-DKO U2OS cells were seeded into a PerkinElmer Cell Carrier Ultra 96 well plate (PerkinElmer) and incubated with 25 µg ml−1 10K-dextran conjugated to Alexa Fluor 647 (D22914; Thermo Fisher Scientific) in DMEM for 12–16 h, and then in fresh DMEM for 2–4 h (chase) to allow dextran to reach the lysosomes3. After that, cells were treated with LLOMe (2 mM, 2 min), cells were then washed within a min and imaged for 40 min after in fresh medium. Images were acquired at 1 min intervals.
Immunofluorescence of lung sections
Tissue sections (see ‘Tissue sectioning’) on SuperFrost Plus Adhesion slides (Fisher Scientific, 11950657) were unfrozen at room temperature and and permeabilized with a 0.5% solution of Tween-20/PBS for 20 min. After that, samples were washed twice with PBS and stained with the corresponding primary antibodies (dilution 1:100) for 2 h. The coverslips were then washed twice with PBS and incubated with the corresponding Alexa Fluor (Thermo Fischer) secondary antibodies (dilution 1:700) for 1 h. After two washes with PBS, the samples were mounted using Dako fluorescence mounting medium (S3023).
Poly(A) RNA in situ protocol
iPSDM cultured in µ-Slide 18-Well Glass Bottom plates (Ibidi) were fixed in 4% PFA for 10 min, PFA aspirated, and 100% cold methanol added to each well for 10 min. Methanol was replaced with 70% ethanol and incubated for 10 min. The ethanol was aspirated and 1 M Tris, pH 8.0 added to each well for 5 min. After Tris removal, hybridization buffer was added containing the dilution of 5′-labelled Cy3-Oligo-dT(20) stock (Integrated DNA Technologies) for a final concentration of 1 ng μl−1. Hybridization was carried out at 37 °C for 2 h. After hybridization, samples were washed once with 4× SSC and then once with 2× SSC (all DEPC-treated). Incubation with primary antibodies was in 2× SSC + 0.1% Triton X-100 for 2 h, washed three times with 2× SSC, and then incubated with secondary antibodies for 1 h at room temperature. Hybridization buffer composition: 1 mg ml−1 Yeast tRNA (AM7119 Thermo Fischer), 0.005% BSA, 10% dextran sulfate (D8906-5G, Merck), 25% Formamide (17899, Thermo Fischer) 20× SSC + DEPC water so that final buffer volume is in 2× SSC.
Samples were fixed by adding a mixture of 8% PFA in 200 mM HEPES buffer to culture medium (v/v) and incubated at room temperature for 15 min then replaced with 4% PFA in 100 mM HEPES for 30 min before imaging by confocal microscope. After imaging by confocal, samples were transferred to 1% glutaraldehyde in 100 mM HEPES buffer.
Fluorescently imaged samples were processed for correlative light and electron microscopy in a Biowave Pro (Pelco) with use of microwave energy and vacuum. Cells were twice washed in HEPES (Sigma-Aldrich H0887) at 250 W for 40 s, post-fixed using a mixture of 2% osmium tetroxide (Taab O011) 1.5% potassium ferricyanide (Taab, P018) (v/v) at equal ratio for 14 min at 100 W power (with/without vacuum 20 inch Hg at 2-min intervals). Samples were washed with distilled water twice on the bench and twice again in the Biowave 250 W for 40 s. Samples were stained with 1 % aqueous uranyl acetate (Agar scientific AGR1260A) in distilled water (w/v) for 14 min at 100 W power (with/without vacuum 20 inch Hg at 2-min intervals) then washed using the same settings as before. Samples were dehydrated using a step-wise acetone series of 50, 75, 90 and 100%, then washed 4 times in absolute acetone at 250 W for 40 s per step. Samples were infiltrated with a dilution series of 25, 50, 75, 100% Durcupan ACM (Sigma-Aldrich 44610) (v/v) resin to propylene oxide. Each step was for 3 min at 250 W power (with or without vacuum 20 inch Hg at 30 s intervals). Samples were then cured for a minimum of 48 h at 60 °C.
Sample trimming and image acquisition
Referring to grid coordinates, the sample block was trimmed, coarsely by a razor blade then finely trimmed using a 35° ultrasonic, oscillating diamond knife (DiATOME, Switzerland) set at a cutting speed of 0.6 mm s−1, a frequency set by automatic mode and a voltage of 6.0 V, on a ultramicrotome EM UC7 (Leica Microsystems, Germany) to remove all excess resin surrounding the region of interest (ROI). Ribbons were cut to a thickness of 65 nm and Images were acquired using a JEOL JEM-1400 series 120 kV transmission electron microscope.
Correlative light and electron microscopy image alignment
Fluorescent images were converted to tiff file format and liner adjustments made to brightness and contrast using FIJI (version 2.9.0/1.53t). Fluorescent images were aligned to serialEM micrographs (TrakEM2) using BigWarp_fiji_7.0.7 plugin. No less than 10 independent fiducials were chosen per alignment for 3D image registration. When the fiducial registration error was greater than the predicted registration error, a non-rigid transformation (a nonlinear transformation based on spline interpolation, after an initial rigid transformation) was applied.
Live-cell imaging analysis of G3BP1+ and GAL-3+ events
An event was considered positive when the fluorescence signal was at least three times greater than the mean background (determined previously to the addition of LLOMe). Intensities were quantified selecting the corresponding ROI in FIJI/ImageJ. G3BP1+ and GAL-3+ events were manually tracked over time and the time registered when the first positive event was detected. At least 20 events per experiment were annotated. The percentage of ‘capping events’ was determined considering the amount of double-positive G3BP1+/GAL-3+ events that presented a polarized G3BP1 fluorescence signal distribution in comparison with the distribution observed for GAL-3+ vesicles. To this end, GAL-3+ areas were segmented and the mean intensity of G3BP1 in those areas determined. A G3BP1+ event was considered polarized (in ‘cap pattern’) when the G3BP1 mean intensity, of a ROI in proximity of a GAL-3+ event, was at least two times greater than the G3BP1 mean intensity corresponding to the segmented (overlapping) GAL-3+ area. For fluorescence intensity or puncta area over time analysis, the corresponding values after segmentation were plotted using the geom_smooth function in R Studio (method = “loess”) (see source data for Fig. 1).
Spatial point pattern analysis
To determine whether the distances between points in group A (G3BP1+) and points in group B (GAL-3+) were significantly different from the distances between points in group A and randomly generated points, we performed a spatial point pattern analysis using the spatstat package in R (version 3.0–6). First, a point pattern object was created from the x and y coordinates of the data using the ppp function. The lambda value, which represents the intensity of the point pattern for group A, was then calculated as the number of points in group A divided by the area of the point pattern. Next, a random point pattern with the same window as the original point pattern was generated using the runifpoint function. The distances between points in group A and points in group B were then calculated using the nncross function, and the distances between points in group A and the randomly generated points were calculated in a similar way. The number of simulated distances that were less than or equal to the observed distances was counted and used to calculate a P value using a Monte Carlo simulation. A total of 1,000 simulations were performed to estimate the P value. Finally, a boxplot was created to visually compare the distribution of observed distances between points in group A and points in group B to the distribution of distances between points in group A and randomly generated points.
Stress granules and GAL-3 puncta analysis
Analysis was done in Fiji/ImageJ using the sequence Image>Adjust>Threshold and then puncta or area in the segmented image was determined using the menu command Analyze>Analyze particles. Size was restricted to particles greater than 0.1 um and the circularity restricted to values between 0.4 and 1.
VT-iSIM images were analysed using the EzColocalization plugin63 on FIJI/ImageJ and Spearman’s rank correlation coefficient (SRCC) values used for quantification. Analysis was done on a single z-stack section of 150 nm.
High-content imaging and evaluation of lysosomal activity and content
iPSDM were incubated with 1 µM solution of the iABP Smart Cathepsin Imaging Probe (40200-100, Vergent Bioscience) for 3 h at 37 °C and 5% CO2. Cells were single-cell segmented based on nuclear staining and lysosomes segmented using the Find Spots and Morphology Properties modules of Harmony 4.9 software. After that lysosomal intensity values (650–760 em) and cellular morphology parameters were quantified.
LysoTracker recovery analysis
iPSDM were segmented using the NucGreen nuclear signal and Find Nuclei module or based on cellular segmentation after applying the Inverse Grey palette and Find Cell module of Harmony 4.9 software. LysoTracker puncta were identified using the Spots building block (local maximum). The spot intensity per cell and timepoint was normalized to the average of relative spot intensity before LLOMe addition. For each experiment, at least 300 cells were analysed per condition.
Mtb replication analysis
Images of Mtb-infected iPSDM were acquired on an Opera Phenix microscope using a 40× objective with at least 20 fields of view per well (with three wells per condition) and analyzed in Harmony 4.9. Cells were segmented based on DAPI, excluding any cells touching the edge of the imaged area. Bacteria were detected using the Find Spots building block of Harmony. The total bacterial area in each cell was then determined. Data was exported and analysed in R Studio to calculate the mean Mtb area per cell for each condition at each timepoint, with all three wells pooled. At least three independent experiments were done per condition and timepoint.
Mouse aerosol infection with Mtb H37Rv
Six- to eight-week-old, female C3HeB/FeJ mice were infected with either wild-type Mtb Wasabi HygR or Mtb ΔRD1 E2-Crimson KanR. Sample size was determined in accordance with ARRIVE guidelines and previous studies64. Five animals per group were used per time of infection. Females were used for safety and space allocation restrictions as infected mice were contained in BSL3. All mice were maintained in BSL3 cages, at 22 ± 2 °C and a relative humidity of 55 ± 10%. No randomization or blinding was applied for this study. For low dose aerosol infection experiments all bacteria were used at mid-exponential phase and a Glascol aerosol generator was calibrated to deliver approximately 100 colony-forming units (CFU) of wild-type Mtb Wasabi HygR/lung and 500 CFU of Mtb ΔRD1 E2-Crimson KanR/lung. Fifty-six days after infection, lungs were perfused in 4% PFA for fluorescence microscopy analysis. Lungs were equilibrated in 0.1 M HEPEs buffer (pH 7.4) with 0.2 M sucrose for 1 h before being transferred to silicon moulds containing OCT medium (Agar scientific, AGR1180). Moulds containing OCT and lungs were then transferred to dry ice and frozen in preparation for sectioning. Sections were cut using a Leica CM30505S Cryostat (CT-18 °C, OT-20 °C), to a size of 8 μm. Sections were collected on SuperFrost Plus Adhesion slides (Fisher Scientific, 11950657) and stored at −80 °C before further processing.
Statistical analysis was performed using GraphPad Prism 10 software or R Studio 2023.03.0 (R version 4.2.2). High-content imaging analysis and mean values were obtained using R 4.2.2 or Harmony 4.9 software. The number of biological replicates, the statistical analysis performed, and post hoc tests used are mentioned in the figure legends. The statistical significance of data is denoted on graphs by informing the P value or asterisks, where *P < 0.05, **P < 0.01, ***P < 0.001; or NS, not significant. Graphs were plotted in GraphPad Prism software or using R Studio 2023.03.0 (R version 4.2.2). RNA-sequencing data were obtained from a previous study42 and plotted using Morpheus (https://software.broadinstitute.org/morpheus/). Schematics were created with BioRender.com.
This study involved the use of KOLF2 human iPS cells, Public Health England Culture Collections, catalogue number 77650100, and the use of WTC human mEGFP-tagged G3BP1 iPS cells (Coriell Institute, AICS-0082 cl.1). The use of human cells is covered and approved by the Ethical Committee and regulated by the Francis Crick Institute Biological Safety Code of Practice in the project registered at the Crick (Project HTA17) framed under Human Tissue Authority Licence number 12650.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.