Post-translational modification-centric base editor screens to assess phosphorylation site functionality in high throughput

Abstract

Signaling pathways that drive gene expression are typically depicted as having a dozen or so landmark phosphorylation and transcriptional events. In reality, thousands of dynamic post-translational modifications (PTMs) orchestrate nearly every cellular function, and we lack technologies to find causal links between these vast biochemical pathways and genetic circuits at scale. Here we describe the high-throughput, functional assessment of phosphorylation sites through the development of PTM-centric base editing coupled to phenotypic screens, directed by temporally resolved phosphoproteomics. Using T cell activation as a model, we observe hundreds of unstudied phosphorylation sites that modulate NFAT transcriptional activity. We identify the phosphorylation-mediated nuclear localization of PHLPP1, which promotes NFAT but inhibits NFκB activity. We also find that specific phosphosite mutants can alter gene expression in subtle yet distinct patterns, demonstrating the potential for fine-tuning transcriptional responses. Overall, base editor screening of PTM sites provides a powerful platform to dissect PTM function within signaling pathways.

Extended Data Fig. 1. Extended Data Figure 1 associated with Figure 1 Kennedy et al.

Phosphoproteomics quality controls. A) Principal component analysis of all phosphoproteomics samples prior to differential expression analysis. B) Multi-scatter plot comparing all samples to each other pairwise. Pearson’s r is shown. Colors of samples are the same as in Extended Fig. 1A.

Extended Data Fig. 2. Extended Data Figure 2 associated with Figure 2 Kennedy et al.

Gating strategy for CD69 staining analyzed by flow cytometry.

Extended Data Fig. 3. Extended Data Figure 3 associated with Figure 3 Kennedy et al.

Gene ontology analysis of the genes targeted by the sgRNAs depicted in Figure 3D. Colors coordinate with Figure 3D. Dotted line is the hypergeometric distribution test FDR threshold.

Extended Data Fig. 4. Extended Data Figure 4 associated with Figure 4 Kennedy et al.

Quality control and characterization of phosphosite base editing coupled to NFAT activity reporters. A) Pairwise Spearman correlations between all normalized log transformed read counts across replicates and experimental conditions. 0.4 is the lower limit cut off in black. B) Mean (across replicates) sgRNA counts for individual sgRNAs prior to collapsing redundant phosphosite targets in the GFP high and low bins. Regression line is shown. C) Percentage of phosphosite targets with one or more protospacer sequences. D) g:Profiler (gene-centric) analysis of genes with phosphosite mutations enriched in the GFP low or GFP high bins. For the x-axis the normalized enrichment score (NES) was multiplied by the -log10 FDR. E) GSEA (gene-centric) analysis of gene sets enriched in the GFP high bin. TCR Calcium Pathway is bolded. F) Proportion of phosphosite targets that contain a putative bystander edit in the library as a whole and in the sorted GFP bins. Student’s two sample T test P value is shown. G) Scatterplot comparing the F statistic from the phosphoproteomic analysis, a proxy for magnitude and reproducibility of abundance changes across the four time points, and the log2 fold change GFPhigh/low bins calculated by MAGeCK. Horizontal red dashed line delineates nominal p value of < 0.05 from the moderated F test of the phosphoproteomics data. H) Scatterplot comparing the log2 fold change GFPhigh/low bins calculated by MAGeCK to the predicted functional score from the machine learning analysis in Ochoa et al. Inset shows the full data structure while the scatter plot is a zoom of points above a predicted functional score of 0.5. Horizontal red dashed line delineates a score threshold determined in Ochoa et al. I) Distribution of predicted functional scores from Ochoa et al. for all data points in the GFP screen, the phosphosite mutants that increased (“up” in red) or decreased GFP levels (“down” in purple). P values for comparison to the whole data set are shown. Data points represent the mean log2 FC (GFPhigh/GFP low) of four transduction replicates. P values for an ANOVA test followed by uncorrected Fisher’s least significant difference for multiple comparisons.

Extended Data Fig. 5. Extended Data Figure 5 associated with Figure 5, Kennedy et al.

EditR software analysis plots outlining bystander base editing levels for PHLPP1 S118P and MAPK1 Y187C prior to single cell clone isolation.

Extended Data Fig. 6. Extended Data Figure 6 associated with Figure 6, Kennedy et al.

A) Log2 fold change of select T cell genes differentially expressed between PHLPP1 S118P and MAPK1 Y187C mutant cells, compared to HEK3 control cells. B) Gating strategy for intracellular GZMB staining and analysis by flow cytometry.

Figure 1

Signaling dynamics of early T cell activation. A) Diagram of early T cell activation and where the time points for the phosphoproteomic analysis were sampled. Pink ovals represent kinases, colored circles are transcription factors. B) Heatmap of statistically regulated (moderated F test) phosphopeptides during the first 27 minutes of T cell stimulation. Interactive data is associated with this figure (Supp Data File 1). C) Heatmap of PTM-SEA terms for the phosphoproteomics time series data. “iKiP” indicates the In Vitro Kinase-to-Phosphosite Database, “pKS” is kinase-substrate pairs from PhosphositePlus (PSP). “PERT” is a PSP-curated perturbation.

Figure 2

Base editing capabilities of empirically derived phosphorylation sites. A) The bar graph shows the number of sgRNAs that target phosphosites for the respective base editors BE4 (pink) or ABE8e (blue) and what the resultant amino acid codon will be for BE4 or ABE8e. BE4 cannot mutate Y to any other amino acid and was omitted. The Venn diagram inset shows the number of distinct phosphosites targeted by at least one sgRNA. B) The distribution of all putative phosphorylated amino acid side chains from all phosphopeptides detected by mass spectrometry, the number of statistically regulated phosphopeptides in Figure 1B, and the number of sgRNAs that can be used by ABE8e to target a phosphosite. C) Diagram for testing various base editor delivery methods in an arrayed format, one sgRNA at a time followed by T cell activation. D) Base editing efficiency, as determined by NGS amplicon sequencing in percent of edited reads, testing the nucleofection of different biomolecules to deliver base editors. The base editing window, counting right to left from the PAM sequence, is shown in the inset and the nucleic acid targets are color coordinated. “p” indicates plasmid, “r” is recombinantly expressed and purified, “mRNA” is synthetic, capped mRNA. E) Effect of phosphosite base edits on T cell activation-induced CD69 surface levels. “Edits” indicates the targeted gene, “Stim” indicates 12 hours of treatment with α-CD3/CD28 agonist antibodies, and “Target” indicates amino acid targeted. NTC, non-targeting control at the HEK3 locus; MFI, mean fluorescence intensity. Two sample T test p-values are shown where “ns” denotes not significant. n = two or three editing replicates, indicated by the data points where standard deviation is shown.

Figure 3

Base editing screening reveals phosphosites involved in proliferation or survival. A) Diagram of pooled base editor screens for phosphosites or terminating edits of essential genes important for cell proliferation or survival followed by next-generation sequencing (NGS). B) Log2 fold change between pre- and post-ABE8e protein introduction of cells expressing the sgRNA library. “Essential” refers to a terminating edit into essential genes. Intergenic base edits and non-targeting controls are also shown. Two sample T test p values are shown, n= 4 transduction replicates. C) Mutations made to CDK1 phosphosites and their influence of sgRNA representation before and after base editing in a pooled format. “sg1 and sg2” indicate that two different sgRNAs were used for the Y15C mutation. Two sample T-test p values are shown, n= 4 transduction replicates where standard deviation is shown. Box and whiskers plot shows median, quartiles, max and min, and outliers (individual data points). D) Volcano plot showing the distribution of all sgRNAs pre- and post-ABE8e protein introduction as determined by MAGeCK analysis. Green points indicate terminating edits in essential genes. Red points indicate statistically significant sgRNA targeting phosphosites depleted after base editing. Blue points are sgRNAs targeting phosphosites that were enriched after base editing. Enrichment values and statistical thresholds were determined by MAGeCK. E) Kinase Library, site-centric enrichment analysis of phosphosite mutants, as an aggregate motif, enriched in post ABE8e-edited cells (blue) or pre ABE8e-edited cells (red) bins. Enrichment Values by MAGeCK and the one-sided Fisher’s exact text p value, after Benjamini-Hochberg, correction is show.

Figure 4

Proteome-wide base editing of phosphosites modulating NFAT transcriptional activity. A) Diagram of PTM-centric, proteome-wide base editing coupled to NFAT-GFP transcriptional reporter followed by next-generation sequencing (NGS). B) Volcano plot comparing phosphosite edits in GFP high (red) compared to GFP low (purple) bins as determined by MAGeCK. Statistical thresholds were also determined by MAGeCK. C) MAGeCKFlute gene-centric pathway analysis of genes with mutated phosphosites enriched in the GFP low (purple) or GFP high (red) bins. Genes in the respective pathways and their fold change are shown. D) Kinase Library, site-centric enrichment analysis of phosphosite mutants enriched in the GFP low (purple) or GFP high (red) bins. Enrichment Values by MAGeCK and the one-sided Fisher’s exact text p value, after Benjamini-Hochberg, correction is show. E) PTM-set Enrichment Analysis (SEA) of phosphosites mutated by ABE8e protein and enriched in the GFP low (purple) or GFP high (red) bins. “iKiP” indicates the In Vitro Kinase-to-Phosphosite Database, “pKS” is kinase-substrate pairs from PhosphositePlus (PSP). MAPK14 is also known as P38A.

Figure 5

Phosphorylation-induced nuclear translocation of PHLPP1 promotes NFAT and represses NFκB transcriptional responses. A) Distribution of log2 fold changes of the ~11,000 sgRNAs inducing phosphorylation mutations in the NFAT (GFP) transcriptional activity screen (top, green) and non-targeting and intergenic controls (lower panel, gray) between GFP low and GFP high bins. PHLPP1 S118P and MAPK1 Y187C mutations are labeled for comparison. B) Validation of NFAT-activity screening hits using electroporation of in vitro transcribed sgRNAs coupled with ABE8e protein, followed by α-CD3/CD28 stimulation for 16 hours and analysis of GFP (NFAT) or CFP (NFκB) transcriptional activity reporters. Two sample T test p values compared to HEK3 relative color control are shown if > 0.001, n=3 activation replicates, where standard deviation is shown. C) Heatmap of differentially expressed genes between HEK3 intergenic mutant control, MAPK1 Y187C, or PHLPP1 S118P edited Jurkat cells activated for 0 or 6 hours as determined by single cell RNA sequencing. The LCP2 terminating edit is shown for comparison but was not used for statistical testing. (Supp Data File 2). D) K-means clustering and g:Profiler gene ontology analysis (all p < 0.05) of the differentially expressed gene clusters at 6 hours between HEK3 intergenic mutant control, MAPK1 Y187C, or PHLPP1 S118. The LCP2 terminating edit is shown for comparison but was not used for statistical testing. Cluster numbers count from left to right and are designated by color. E) Spinning disk confocal microscopy images showing the subcellular localization of PHLPP1 N-terminal extension (NTE) constructs. “NLS mut” refers to full mutation of the two nuclear localization sequences in PHLPP1’s NTE. The S118P, S118A and S118E mutations are also shown. F) Quantification of PHLPP1 NTE constructs and the percentage of α-HA signal in the nucleus. Unpaired, two sample T test p values are shown. Each data point represents an individual cell, where standard deviation is plotted.

Figure 6

Dissecting T cell activation transcriptional responses. A) Select T cell-related differentially expressed genes at six hours post T cell activation between HEK3 (control), PHLPP1 S118P, and MAPK1 Y187C mutant cells. This data in log2 fold change space is available as Supp. Data file 2. B) Intracellular GZMB staining in HEK3 (control), PHLPP1 S118P, MAPK1 Y187C or LCP2 term edited mutant cells 24 hours post-T cell activation. Representative plot (left) and quantification of % GZMB⁺ cells, n = 6 activation replicates where standard deviation is shown.