Identification of phagocytosis regulators using magnetic genome-wide CRISPR screens

Abstract

Phagocytosis is required for a broad range of physiological functions, from pathogen defense to tissue homeostasis, but the mechanisms required for phagocytosis of diverse substrates remain incompletely understood. Here, we developed a rapid magnet-based phenotypic screening strategy, and performed eight genome-wide CRISPR screens in human cells to identify genes regulating phagocytosis of distinct substrates. After validating select hits in focused miniscreens, orthogonal assays and primary human macrophages, we show that (1) the previously uncharacterized gene NHLRC2 is a central player in phagocytosis, regulating RhoA-Rac1 signaling cascades that control actin polymerization and filopodia formation, (2) very-long-chain fatty acids are essential for efficient phagocytosis of certain substrates and (3) the previously uncharacterized Alzheimer's disease-associated gene TM2D3 can preferentially influence uptake of amyloid-β aggregates. These findings illuminate new regulators and core principles of phagocytosis, and more generally establish an efficient method for unbiased identification of cellular uptake mechanisms across diverse physiological and pathological contexts.

Figure 1.. Genome-wide CRISPR screening for genetic regulators of phagocytosis.

a. Schematic of phagocytosis screening strategy using magnetic separation. Labeling with ferrous nano-particles enables magnetic separation screening of diverse substrates, and comparison with paramagnetic beads of different size and charge. b. Confocal microscopy demonstrates efficient magnetic separation of phagocytosing and non- phagocytosing cells from a mixed population (1.3 μm magnetic beads). Representative of 4 independent experiments. c. sgRNA distributions for two example hits and the distribution of negative control sgRNAs from a genome-wide phagocytosis screen using 1.3 μm magnetic beads. A positive value indicates enrichment in the magnet-bound fraction, and a negative value indicates enrichment in the unbound fraction. Enrichment of 10 sgRNAs targeting NCKAP1L and CDK2 are represented with blue and red vertical lines, respectively. The distribution ~10,000 negative controls is shown in black. d. Volcano plot of all genes indicating the effect and confidence scores for a genome-wide phagocytosis screen using 1.3 μm magnetic beads. Effect and confidence scores are determined by casTLE. The 85 genes passing 10% FDR for inhibiting or promoting phagocytosis are highlighted in blue and red respectively. e. Select Reactome categories enriched in the 85 genes that pass the 10% FDR cutoff as determined by casTLE (Reactome category IDs listed in methods – some categories are abbreviated). f. Phagocytosis of diverse substrates is actin dependent. Cells were pretreated with the actin polymerization inhibitor cytochalasin D (CytoD) before exposure to ferrous nano-particle-labeled substrates, and binding to the magnetic column was compared to uninhibited cells. g. Comparison of phagocytosis screen results with eight different substrates using unbiased hierarchical clustering, based on 150 genes that reached a <5% FDR cutoff as determined by casTLE in at least three screens. Select, functionally related sets are highlighted.

Figure 2.. Putative modifiers of phagocytosis of diverse substrates from eight genome-wide screens.

Bar plots represent the signed log₁₀ of the p-value for each gene for each substrate as calculated by casTLE using two replicate screens for each of eight different substrates. Each gene represented passed a 5% FDR threshold in at least three genome-wide phagocytosis screens. A negative value (blue bars) indicates that the gene knockout inhibits phagocytosis, and a positive value (red bars) indicates that the gene knockout promotes phagocytosis. Annotations of cellular processes and compartments were based on literature and GO-term assignments.

Figure 3.. Orthogonal measurements of phagocytosis validate magnetic screen results.

a. Design of FACS-based validation screen. Differentiated U937 cells containing a custom sgRNA library targeting 322 genes identified in the genome-wide screens were FACS-sorted based on fluorescence signal from pHrodo-labeled zymosan. b. Correlation of casTLE effect scores between FACS and magnetic separation screens using the 322-gene custom sgRNA library. c. U937 cells expressing a NCKAP1L sgRNA treated with pHrodo-labeled zymosan (red). Live cells were labeled with Calcein AM (green). Representative of six independent experiments. Scale bars = 50 μm. d. Automated live cell microscopy was used to monitor signal from pHrodo zymosan over time. A phagocytic index was calculated by measuring the total area of pHrodo signal divided by the total live cell area, then normalized to the average value of the control lines at 5 hr. Values represent mean ± S.E.M of n=4 replicate wells. e. Summary of additional validations. The phagocytic index at 5 hr is presented. Values represent mean ± S.E.M of n=4 replicate wells. (**p< .01, *p< .05, two-way ANOVA with Dunnett’s comparison to control sgRNA). f. Validation in primary human macrophages. FACS measurement of pHrodo-labeled bead phagocytosis with either control sgRNA or sgRNAs targeting NCKAP1L or NHLRC2. Results are normalized to the maximal response by each independent donor. Values represent mean ± S.E.M of n=4 (NHLRC2 or NCKAP1L) or n=6 (Control) technical replicates. Dots represent averaged technical replicates across n= 2 biological donors. **p<.01, ****p<.0001 (one-way ANOVA with multiple-comparisons correction). g. Schematic for competitive phagocytosis assay to validate substrate specific regulators of phagocytosis. h. Example images from competitive phagocytosis assays of zymosan (red) and 1.3 μm beads (green) in U937 cells expressing a control sgRNA, TLN1 sgRNA, or PLEK sgRNA. Images are representative of two independent experiments. Scale bars = 10 μm. i. Ratio of total red area (zymosan) to total green area (bead) at 5 hours in cells expressing control sgRNA or sgRNAs targeting TLN1 or PLEK. Values represent mean ± S.E.M. (*p<.05, two-tailed t- test from n=4 technical replicates; results are representative of three independent experiments). j. Phagocytic index of zymosan (red) and beads (green) in competitive phagocytosis assay over time. Values represent mean ± S.E.M of n=4 replicate wells; results are representative of three independent experiments.

Figure 4.. Loss of NHLRC2 inhibits phagocytosis through Rho/Rac1 signaling.

a. Clonally derived U937 cells with verified NHLRC2 knockout were labeled with Calcein AM (green) after 5 hr of exposure to pHrodo labeled zymosan (red). Representative of six independent experiments. Scale bar = 50 μm. b. Phagocytic index (see Figure 3) of pHrodo labeled zymosan from control sgRNA expressing U937s and clonally derived NHLRC2 KO U937 cells. Values represent mean ± S.E.M. of n=4 replicate wells. c. Results from BiolD for NHLRC2. Enrichment of a protein in the NHLRC2-BirA pulldown versus BirA alone is plotted against the mean confidence score (SAINT score) (n=3 experimental replicates). The bait protein (NHLRC2) is highlighted in red and top protein interactors are highlighted in blue. d. Quantification of active (GTP-bound) RhoA in RAW 264.7 cells expressing a control sgRNA (red) or confirmed Nhlrc2 KO RAW 264.7 cells (blue) as determined by. ELISA signal was normalized against recombinant GTP-bound RhoA. Values represent mean ± S.E.M. of n=3 experimental replicates (**p<.005, two-tailed t-test). e. Phagocytic index measurements using pHrodo labeled zymosan of RAW 264.7 cells expressing a control sgRNA (grey) with constitutively active (CA) RAC1 (black), Nhlrc2 KO cell line (yellow) with active (CA) RAC1 (red), Nckap1l KO cell line (blue) with constitutively active (CA) RAC1 (green). Values represent mean ± S.E.M for n = 4 replicate wells. f. Scanning electron microscopy of RAW 264.7 cells with control sgRNA (top row), Nckap1l KO (middle row), and Nhlrc2 KO (bottom row) with (left) and without beads (right). Representative images of 3 experimental replicates. Scale bar = 2 μm g. F-actin staining (AlexaFluor 488 phalloidin) of RAW 264.7 with control sgRNA (left) and Nhlrc2 KO (right). Representative images of 3 experimental replicates. Scale bars = 10 μm. h. F-actin staining of control sgRNA (left) and Nhlrc2 KO (right) RAW 264.7 cells engaging IgG-coated coverslips during frustrated phagocytosis. Scale bars = 10 μm i. Model for role of NHLRC2 in phagocytosis.

Figure 5.. Role for very long chain fatty acids (VLCFAs) in phagocytosis.

a. Genes identified in the genome-wide screens as required for phagocytosis (ELOVL1, CERS2, and SGMS1) participate in consecutive steps of a lipid biosynthetic pathway. b. Validation of pathway members using automated microscopy. Phagocytosis was monitored over time as in Figure 3 using U937 cells containing the indicated guide and pHrodo-labeled zymosan. Values represent mean ± S.E.M, n=6 replicate wells. c. Rescue of ELOVL1 deficiency by LCFA supplementation. Phagocytosis measured at 5 hr from U937 cells expressing control (grey), ELOVL1 (light blue), or CERS2 sgRNA (dark blue) supplemented with C18 or C24 long chain fatty acids (LCFA). Values represent mean ± S.E.M, n = 6 replicate wells. (**p< .01, one-way ANOVA with Dunnett’s comparison to control sgRNA). d. Phagocytic index over time for U937 lines expressing a control sgRNA (grey), an ELOVL1 sgRNA (blue), or an ELOVL1 sgRNA and supplemented with C24 LCFA (red). Values represent mean ± S.E.M, n=4 replicate wells. e. Validation of ELOVL1 phagocytic function in primary human macrophages. FACS measurement of the pHrodo-labeled zymosan phagocytosis by macrophages electroporated with either control sgRNA or sgRNAs targeting ELOVL1. Results normalized to the maximal response by each independent donor. Values represent mean ± S.E.M of n=5 (Ctrl or sgRNA 2) or n=6 (sgRNA 1) technical replicates. Experiments performed with n=3 biological donors. Dots represent technical replicates.*p<.05, ****p<.0001 (one-way ANOVA with multiple-comparisons correction). f. RAW 264.7 cells expressing a control sgRNA (top) or ELOVL1 targeting sgRNA (bottom) were incubated with IgG-coated 7 μm beads for 10 minutes and fixed. IgG signal (red) marks bead surfaces unobstructed by cellular contact. Cells were stained with phalloidin (green) and Hoechst nuclear stain (blue). White asterisk marks a fully internalized bead. Representative of two independent experiments. Scale bars = 10 μm. g. Scanning electron microscopy of RAW 264.7 cells expressing a control sgRNA (top) or ELOVL1 sgRNA (bottom) that were incubated with IgG-coated 7 μm beads for 10 minutes and fixed. White asterisk marks a fully internalized bead.

Figure 6.. Identification of substrate specific regulators of phagocytosis.

a. Ratio of total red area (indicating phagocytosis of Amyloid β (Aβ) aggregates) to total green area (indicating phagocytosis of synaptosomes) at final time point of phagocytosis assay in cells expressing control sgRNA or confirmed U937 TM2D3 and TM2D2 clonal knockout lines. Values represent mean ± S.E.M. of n=4 replicate wells (*p<.05, two-tailed t-test). b. Competitive phagocytosis assay in U937 cells expressing control sgRNAs or confirmed U937 TM2D3 and TM2D2 clonal knockout lines. Synaptosomes are labeled with pHrodo green and Amyloid β (Aβ) aggregates labeled with pHrodo red. Phagocytic index for each substrate measured over time using automated live cell imaging. Values represent mean ± S.E.M. of n=4 replicate wells (*p<.05, two-tailed t-test). c. Phagocytosis of Amyloid β (Aβ) aggregates (labeled with pHrodo red) is impaired in confirmed clonal U937 TM2D3 knockout line when compare to U937s expressing control sgRNA. Live U937 cells labeled with Calcein AM (green). Representative of three independent experiments. Scale bar = 10 μm. d. Validation in primary human macrophages of role of TM2D3 in Amyloid β (Aβ) aggregates and zymosan phagocytosis. Flow cytometry-based measurement of the phagocytosis of pHrodo-labeled Amyloid β (Aβ) aggregates or of pHrodo-labeled zymosan by genetically modified macrophages electroporated with either control sgRNA or sgRNAs targeting TM2D3. Results normalized to the maximal response by each independent donor. Values represent mean ± S.E.M of n=2–4 technical replicates. Experiments performed with n=2 biological donors. Dots represent technical replicates. **p<.01. (one-way ANOVA with multiple-comparisons correction).