Base editing screens map mutations affecting IFNγ signalling in cancer

IFNγsignalling underpins host responses to infection, inflammation and anti-tumour immunity. Mutations in the IFNγsignalling pathway cause immunological disorders, haematological malignancies, and resistance to immune checkpoint blockade (ICB) in cancer, however the function of most clinically observed variants remain unknown. Here, we systematically investigate the genetic determinants of IFNγresponse in colorectal cancer cells using CRISPR-Cas9 screens and base editing mutagenesis. Deep mutagenesis of JAK1 with cytidine and adenine base editors, combined with pathway-wide screens, reveal loss-of-function and gain-of-function mutations with clinical precedence, including causal variants in haematological malignancies and mutations detected in patients refractory to ICB. We functionally validate variants of uncertain significance in primary tumour organoids, where engineering missense mutations in JAK1 enhanced or reduced sensitivity to autologous tumour-reactive T cells. By classifying > 300 missense variants altering IFNγ pathway activity, we demonstrate the utility of base editing for mutagenesis at scale, and generate a resource to inform genetic diagnosis.


Abstract
IFNg signalling underpins host responses to infection, inflammation and anti-tumour immunity. Mutations in the IFNg signalling pathway cause immunological disorders, haematological malignancies, and resistance to immune checkpoint blockade (ICB) in cancer, however the function of most clinically observed variants remain unknown. Here, we systematically investigate the genetic determinants of IFNg response in colorectal cancer cells using CRISPR-Cas9 screens and base editing mutagenesis. Deep mutagenesis of JAK1 with cytidine and adenine base editors, combined with pathway-wide screens, reveal lossof-function and gain-of-function mutations with clinical precedence, including causal variants in haematological malignancies and mutations detected in patients refractory to ICB. We functionally validate variants of uncertain significance in primary tumour organoids, where engineering missense mutations in JAK1 enhanced or reduced sensitivity to autologous tumour-reactive T cells. By classifying > 300 missense variants altering IFNg pathway activity, we demonstrate the utility of base editing for mutagenesis at scale, and generate a resource to inform genetic diagnosis. Cellular responses to the cytokine interferon g (IFNg) are essential for normal inflammatory responses, but pathway dysfunction and disease can occur through mutation, leading to haematological malignancies and immunological disorders 1,2 . JAK kinase inhibitors are used to treat myeloproliferative disorders such as polycythaemia vera, and inflammatory disorders such as rheumatoid arthritis and ulcerative colitis 2 , reflecting the central role of JAK-STAT signalling in these diseases. Furthermore, IFNg signalling in cancer cells is a critical aspect of anti-tumour immunity 3,4 . Clinical resistance to ICB, such as antibody therapies targeting PD-1 and CTLA-4, has been associated with somatic mutation and homozygous inactivation of IFNg pathway components in tumour cells [5][6][7][8] , or inactivation of genes involved in antigen processing and presentation (e.g. B2M) 9,10 that are expressed in response to IFNg. For example, mutations in JAK1 and JAK2 can confer resistance to ICB 5,6 .
However, such loss-of-function (LOF) mutations in IFNg pathway components are rare, reflecting the limited number of tumour samples sequenced pre-and post-ICB to date 11 , and the apparent absence of convergence (hotspots), which is more common in resistance to small molecule inhibitors 12 . Since somatic mutations in cancer are predominantly single nucleotide changes, which often result in missense mutations with unknown consequence 13,14 (i.e. variants of uncertain significance, or VUS), interpreting their functional relevance remains challenging, representing an impediment to diagnosis, patient stratification, and management of drug-resistant disease.
Experimental approaches are instrumental in assessing the functional effects of VUS. This is due to the ability to establish causality between VUS and disease-related phenotypes, as well as a scarcity of clinical datasets (e.g. from sequencing ICB-resistant tumours), and the infrequent occurrence of some variants in patient cohorts. For example, cDNA-based expression of variant alleles can be used 12 , but this is not easily scaled and does not reflect physiological levels of gene expression. Bioinformatic predictions of variant effect are not completely predictive and often discordant 15 . Saturation genome editing (SGE) using CRISPR-mediated introduction of exogenous homology-directed repair (HDR) templates 15 is challenging to scale to multiple genes, costly, and often limited to cell lines with high levels of HDR and near-haploid genomes, which can restrict its utility for studying VUS in diseaserelevant cell models. Another methodology to prospectively assess endogenous gene variant function at scale is base editing [16][17][18][19][20] ; a CRISPR-based gene editing technology that employs cytidine 21 or adenine 22 deaminases to install C->T or A->G transitions, respectively, achieving high editing efficiencies with minimal generation of DNA insertions and deletions (indels).
In this study, we use CRISPR-Cas9 screening to identify mediators of sensitivity and resistance to IFNg in colorectal adenocarcinoma (CRC), and use cytidine base editors (CBEs) and adenine base editors (ABEs) to perform mutagenesis of the top-scoring genes, thereby systematically mapping LOF and gain-of-function (GOF) variants modulating IFNg pathway activity (Fig. 1a), including VUS associated with diseases such as cancer.

CRISPR-Cas9 screens identify mediators of sensitivity and resistance to IFNg
To systematically evaluate genetic, cell-intrinsic determinants of IFNg signalling, and nominate genes for further investigation, we performed CRISPR-Cas9 screens in two colorectal cancer cell lines, HT-29 and LS-411N (both BRAF mutant, and microsatellite stable and microsatellite unstable, respectively) (Fig. 1a). Cas9-expressing derivative cell lines 23 were transduced with an immuno-oncology focused guide RNA (gRNA) gene knock-out (KO) library, containing 10,595 gRNAs targeting 2,089 genes with a median of five gRNAs per gene (Supplementary Table 1) and selected with cytotoxic doses of IFNg. Screen quality was verified by efficient depletion of gRNAs targeting essential genes 24,25 (Supplementary Fig.   1a), and correlation between independent biological screening replicates ( Supplementary   Fig. 1b).
MAGeCK 26 (Fig. 1b) and Drug-Z 27 ( Supplementary Fig. 1c) analyses indicated that KO of genes involved in the regulation of IFNg signalling, JAK-STAT signalling, and the downstream transcriptional response, caused the strongest resistance, including IFNGR1, IFNGR2, JAK1, JAK2, STAT1 and IRF1 (Fig. 1b), each of which had multiple gRNAs with significant enrichment specifically in the presence of IFNg ( Fig. 1c and Supplementary Fig.   1d). Changes in gRNA abundance were generally greater for HT-29, reflecting higher sensitivity to IFNg and a faster growth rate than LS-411N ( Supplementary Fig. 1e).
Identification of hits common to both cell lines (Fig. 1d) and STRING network analysis 28 revealed genes centred around IFNg signalling, protein ubiquitination, RNA processing, and mTOR signalling ( Supplementary Fig. 1f). , AKT1 and WDR24 were significantly associated with resistance to IFNg,   whereas negative regulators of mTOR, TSC1 and STK11, were sensitising hits, consistent with the pleiotropic, immunosuppressive effects of rapamycin, and mTOR signalling potentiating IFNg signalling 29 . Gene function enrichment analysis 30 suggested sensitising and resistance hits were highly enriched for ubiquitin mediated proteolysis and antigen processing pathways 10 (Fig. 1b, Supplementary Fig. 1g). Inactivation of genes involved in protein degradation such as tumour suppressor genes KEAP1 and FBXW7, have been previously implicated in sensitivity and resistance to cancer immunotherapy, respectively 31,32 . Interestingly, FBXW7 was a significant resistant hit in HT-29 but not LS-411N, where FBXW7 is already mutated 33 . Moreover, sensitising hits included KO of SOCS1 and STUB1 34 , which are negative regulators of IFNg signalling that function through inhibition and proteasomal degradation of JAK1 35  Our CRISPR-Cas9 screens identified key nodes of resistance and sensitivity to IFNg in CRC cell lines for further study, with considerable overlap with clinical reports of ICB resistance in patients [5][6][7] , and genetic screens interrogating cancer immune evasion in vitro 10,31,36,37 and in vivo 34,36,38 (Supplementary Discussion, Supplementary Table 1).

Figure 1. CRISPR-Cas9 screens identify mediators of IFNg sensitivity and resistance a)
Schematic of the integrated CRISPR-Cas9 and base editing screening approaches to identify genetic mediators of sensitivity and resistance to IFNg. Cas9 was used to identify important pathways and genes regulating IFNg response in colorectal cancer cell lines. Multiple base editing mutagenesis screens were used to assess the functional consequence of variants of uncertain significance (VUS) in key regulators. b) Gene-level volcano plots of CRISPR-Cas9 screens comparing IFNg-treated to control conditions. Data are the average from two independent screens. c) gRNA-level analysis of top resistance genes, representing essential components of the IFNg pathway. d) Common and private genes conferring sensitivity and resistance to IFNg in HT-29 and LS-411N CRC cell lines identified from CRISPR-Cas9 screens.  Fig. 2b), but found that MLH1 was dispensable for base editing in this context ( Supplementary Fig. 2c).
Using a pooled library of 2,000 gRNAs, we tiled JAK1 in HT-29 iBE3 cells with 665 exon-targeting gRNAs and gRNAs targeting JAK1 promoter regions, non-targeting (NT), intergenic targeting, and controls gRNAs designed to introduce stop codons in 72 essential and 28 non-essential genes (Supplementary Table 2). We adopted two screening approaches; a long-term proliferation screen, and a short-term flow cytometry-based assay, based on MHC-I and PD-L1 induction with IFNg (Fig. 2c). gRNAs predicted to cause stop codons within essential genes were significantly depleted (Fig. 2d), achieving recovery of known essential genes in both screens (AUC = 0.65; Supplementary Fig. 2d). There was no relationship between gRNA functional scores and the number of off-target sites 40 ( Supplementary Fig. 2e), however, the gRNA Rule Set 2 score 41 (P = 9.0x10 -4 ; Supplementary   Fig. 2f), or considering the immediate sequence context of the target cytidine 21 ( Supplementary Fig. 2g), was somewhat predictive of gRNA performance 19,20 Fig. 2f; validation cohort). In addition, we included JAK1 Glu890 gRNA, which was unusual as it scored in the proliferation screens but not the FACS screens (Fig. 2f), and the Trp690* gRNA as a control; predicted to generate a nonsense mutation observed in a CRC patient that failed to respond to ICB 6 . shift. Data are representative of two independent experiments performed on separate days. c) FACS screening assay. After base editing of JAK1 by the addition of doxycycline, HT-29 iBE3 cells that failed to respond to IFNg after 48 h were selected by FACS, as determined by lack of induction of MHC-I and PD-L1 expression. Data are representative of two independent experiments performed on separate days. d) Proliferation screening assay. gRNA depletion or enrichment is indicated by z-score, comparing control arm to T0 (time 0) control sample. Base editing gRNAs designed to introduce stop codons in essential genes in HT-29 iBE3 cells are depleted. e) Correlation between screening replicates and different assays. z-scores for gRNAs targeting JAK1 were compared between replicates and alternative screening assays, with each replicate representing an independent screen performed on a separate day. The shaded line area represents the 95 % confidence interval. f) Identification of LOF and GOF alleles in JAK1 protein affecting sensitivity to IFNg. zscores for the base editing screens using FACS vs proliferation were plotted to robustly select potential LOF (blue) and GOF (red) JAK1 variants. Labelling illustrates amino acid positions that were selected for further validation.

Base editing mutagenesis of the IFNg pathway
Having established a robust base editing system, to achieve a more comprehensive overview of functional missense mutations in the IFNg pathway, we expanded our base editor mutagenesis screens to include top hits of our CRISPR-Cas9 screen using HT-29 iBE3-NGG ( Fig. 1b). We tiled JAK1, JAK2, IFNGR1, IFNGR2, STAT1, IRF1, B2M and SOCS1 with 4,608 gRNAs, including the previous JAK1 gRNAs to serve as internal controls (Fig. 3a). B2M was included because of its role in MHC-I presentation and anti-tumour immunity, but it was not a hit in our initial IFNg survival screens as B2M variants should not have an effect on cell proliferation in vitro.
Proliferation and FACS screens were significantly correlated (R 2 adj. 0.42), as were independent replicate screens (proliferation R 2 adj. 0.37; FACS R 2 adj. 0.34; Supplementary   Fig. 3b), each displaying a high level of enrichment of gRNAs predicted to introduce splice variants, stop codons and start-lost mutations (Fig. 3b). Once again, JAK1 Glu890 gRNA was enriched in the proliferation screen, but not in the FACS screen. Such behaviour was rare for most proteins except for the transcription factor STAT1, where a cluster of LOF missense mutations was enriched only in the proliferation screen (Fig. 3b), possibly indicating separation-of-function mutants. Encouragingly, we recovered validated gRNAs targeting JAK1 in this larger screen (Supplementary Table 3, and later sections). In addition to protein truncating mutations, we used JAK1 LOF and GOF gRNAs from our validation cohort as a benchmark for setting the thresholds to call high-confidence functional missense variants in the IFNg pathway (Fig. 3b).   Table 4). For comparison, we included JAK1 screening data from pathway-wide base editing screens. For NGN base editors, we detected significantly enriched gRNAs utilising all four PAMs ( Supplementary Fig. 4a).
ABE cannot introduce stop codons, but predicted splice variants in JAK1, which could be introduced with both CBE and ABE, were significantly enriched over NT control gRNAs in all screens ( Supplementary Fig. 4b). Given the PAM utility and editing windows of each base editor, we predicted non-synonymous amino acid mutation coverage of JAK1 was improved to approximately 39.6 % for BE4max-YE1-NGN, 50.8 % for BE3.9max-NGN, 64.9 % for ABE8e-NGN, and 85.1 % when combining cytidine and adenine NGN mutagenesis. However, we cannot guarantee the editing efficiency of all gRNAs, so the absence of a significant score cannot be used as evidence for the lack of function of an amino acid position.

Deep mutagenesis of JAK1 reveals LOF and GOF variants with clinical precedence
To aid interpretation of our mutagenesis screens, we compiled a database of variants from COSMIC 13  Val658, is mutated in acute myeloid leukaemia (AML); this residue is structurally analogous to JAK2 Val617, which is commonly mutated in polycythaemia vera 1,2 . CBE and ABE screens converged on a cluster of GOF variants in the C-terminus of the kinase domain (Met1099, Arg1103) in a known protein-protein interaction motif for SOCS1 35 (Fig. 4a); a significant negative regulator in our CRISPR-Cas9 screens. These variants presumably disrupt this interaction, increasing JAK1 protein abundance and activity (Fig. 4d). Indeed, amplification of SOCS1 has been found in patients that failed to respond to ICB 7 , implying this regulatory mechanism is of clinical relevance.
LOF positions included Gly887 (Fig. 4a), which is within the kinase active site, with the crystal structure 35 suggesting mutation of this residue would negatively affect Mg 2+ and ATP/ADP coordination (Fig. 4d). Other LOF mutations involving kinase catalytic residues included Asp1003 (proton acceptor), and Asp1021 (within the DFG motif), which were detected with increased (NGN) saturation (Fig. 4a). ABE screens were more likely to detect sites of post-translational modification due to the ability to modify tyrosine, threonine and serine (phosphorylated) and lysine (ubiquitinated), revealing Tyr993, and the known activating Tyr1034 phosphosite as candidate LOF positions (Fig. 4a) IFNg signalling (Fig. 3a), many of which had previously unknown function (Supplementary Table 5). Crystal structure (6C7Y) shows catalytic LOF mutations (blue) proximal to the ATP/ADP binding pocket in the kinase domain, and GOF mutations (red) in the binding interface with the negative regulator SOCS1. e) Western blotting analysis of HEK293T cells following overexpression of FLAG-tagged WT or Gly590Arg mutant JAK1, with or without IFNg stimulation for 1 h. Reduced p-STAT1 signalling is independently replicated in Supplementary Fig. 5d.

Functional validation of variants conferring altered sensitivity to IFNg
We set out to functionally validate 24 gRNAs comprising our JAK1 validation cohort ( Fig. 2f) in an arrayed format, with multiple assays assessing cell proliferation, signalling, protein expression and RNA expression ( Fig. 5a and Fig. 5b). This analysis was germane to screening results from multiple base editing modalities, due to their convergence on JAK1 residues within the validation cohort (e.g. Arg108, Gly590, Asp775, Gly887, Met1099; Figure   4a and Supplementary Table 5). The growth of HT-29 iBE3 cells with engineered JAK1 variants in the presence of IFNg tracked with screen z-scores, with GOF variants having no survival benefit and LOF variants having robust resistance to IFNg, relative to controls (Fig.   5a, Supplementary Fig. 5d). Base editing screening gRNAs were validated, with the possible exception of JAK1 Glu890, which scored poorly in the FACS screen (Fig. 2f), highlighting the value of implementing two screening assays.
Many of the candidate LOF variants had reduced levels of pSTAT1 induction, whereas GOF variants had enhanced levels of pSTAT1 (Fig. 5b, Supplementary Fig. 5d).
Met1099 and Arg1103 GOF variants had increased levels of JAK1 protein and JAK-STAT signalling, consistent with disruption of the SOCS1 binding interface and reduced E3 ubiquitin ligase-mediated destruction 35 . Surprisingly, the Gly590 LOF variants also had elevated levels of JAK1 protein (Fig. 5b, Supplementary Fig. 5d), despite reduced sensitivity to IFNg in terms of cell proliferation and signalling. We speculated that increased JAK1 Gly590Arg protein could also be attributable to altered binding to SOCS1, however, we did not observe any change in binding in co-immunoprecipitation experiments ( Supplementary   Fig. 5e). JAK1 706/707 gRNA targets a splice region and had severely reduced JAK1 protein expression similar to the clinical Trp690* nonsense control (Fig. 5b). The Glu1123 splice variant reduced JAK1 RNA abundance to levels comparable to the Trp690* nonsense control, which we presumed was targeted for nonsense mediated decay. However, basal In sum, these data represent a comprehensive profile of base editing outcomes at endogenous DNA loci, and demonstrates the predictability and precision with which functional variants can be installed. We note that the specificity of editing is retained under strong positive selection pressure, which may be an advantage of transient expression of base editors from a doxycycline-inducible system.

HT55 (CRC) and K2 (melanoma) cell lines harboured homozygous Glu1051Gln and Ala760Val
putative JAK1 LOF missense mutations, respectively ( Supplementary Fig. 7a). As predicted, HT55 and K2 failed to respond to IFNg compared to JAK1 WT cancer cell lines, as measured by failure to induce MHC-I and PD-L1 expression (Fig. 6a). The endogenous C->T mutation in K2 cells was amenable to correction by adenine base editing. ABE8e-NGN-mediated reversion of this JAK1 mutation led to restoration of response to IFNg ( Supplementary Fig.   7b), verifying that this variant is responsible for resistance to IFNg. These data indicate that our base editing variant map is of broad utility, and not private to a particular cell model or tissue type. Interestingly, most of these cancer cell lines were derived before ICB was widely available, which suggests these variants arose from in vivo immunoediting 3,9 rather than acquired therapy resistance.
To assess the relevance of our findings in a more translational setting, we applied base editing to a primary tumour organoid (CRC-9, harbouring FBXW7 and TP53 driver mutations) derived from an MSI colorectal cancer patient where autologous, tumourreactive T cells have been derived from the patient's PBMCs 59,60 (Fig. 6b). Following enrichment for tumour reactive populations and expansion, co-cultured PBMCs were exclusively CD3 + , implying a high proportion of T cells 59,60 ( Supplementary Fig. 7c). Firstly, we confirmed that base editing of JAK1 to install clinically observed missense variants in CRC-9 tumour organoids altered sensitivity to IFNg, as measured by cell proliferation in 3D, with LOF mutations conferring resistance and the GOF mutation JAK1 Met1099Ile increasing sensitivity ( Supplementary Fig. 7c). Next, we used a co-culture of matched tumour-reactive T cells with genetically engineered tumour organoids to assess T cell mediated killing by flow cytometry (Fig. 6c). In this setting, T cell mediated killing of organoids was dependent on MHC-I, pre-exposure of organoids to IFNg to increase MHC-I expression and antigen presentation, but not PD-1 inhibition with nivolumab, or CD28 co-stimulation ( Supplementary Fig. 7d). Strikingly, all JAK1 LOF mutant tumour organoids had significant resistance to anti-tumour T cell mediated killing relative to WT controls, with some mutants achieving survival comparable to antibody blockade of MHC-I, or growing tumour organoids in the absence of T cells (Fig. 6d). Conversely, the GOF mutant Met1099Ile had increased sensitivity to T cell mediated attack.
Taken together, these data illustrate that IFNg-pathway variant maps from base editing screens may be prognostic of anti-tumour immunity. Our data also highlights that JAK1 GOF can sensitise immuno-resistant FBXW7-mutant cancers 32 to T cells. Counting beads were used to quantify absolute cell counts. Data are representative of three biological replicates. d) Quantification of T cell-mediated killing of autologous tumour organoids from flow cytometry analysis. Data represent the average ± SD of three biological replicates, and were compared against parental co-culture controls using an unpaired, twotailed Student's t-test (P ** <0.01, *<0.05). NT, non-targeting gRNA; ø par., parental tumour organoid.

Discussion
In this report, we perform a total of 18 screens with CRISPR-Cas9 and base editors to systematically catalogue the genetic dependencies of IFNg response in CRC cells, and map > 300 missense mutations affecting IFNg pathway activity (see also Supplementary Discussion). Through the use of multiple cytidine and adenine base editors, to the best of our knowledge, this study represents one of the most saturating base editing mutagenesis screens performed to date 19,20,61 . Furthermore, we deploy base editors to systematically study protein structure and function throughout a signalling pathway. We provide BE-view as an online resource to facilitate exploration of these data: www.sanger.ac.uk/tool/beview.
Tumour cell sensitivity to IFNg is an important determinant of ICB response in multiple tumour types [5][6][7][8] . JAK1 is mutated in approximately 10 % of CRC and 6 % of skin cutaneous melanoma, with a significant decrease in survival for melanoma patients with deleterious JAK1 alterations 6 . We detected known LOF variants (JAK1 Asp775Asn, Trp690*) 6 and assigned LOF to VUS in JAK1 that may have contributed to primary or acquired resistance to ICB resistance in the clinic (e.g. JAK1 Gly590Arg, Gly182Glu, Gly655Asp, Pro674Ser) 54,56 . We also discovered a splice mutation in JAK1 as a high-confidence LOF variant (Arg110 splice variant), however this LOF mutation was recorded in a patient's tumour with a partial response to anti-CTLA-4 54 . This highlights that the presence or absence of LOF variants in the IFNg pathway in a tumour biopsy is not an absolute determinant of ICB response; rather, outcome is dependent on multiple factors including the penetrance of the mutation itself (i.e. zygosity), tumour clonal architecture, co-occurring mutations, tumour mutational burden, oncogenic signalling, tumour microenvironment, antigen presentation and immune checkpoint engagement 4,11 . Further work is required to establish the relative importance of each of these determinants, which will be increasingly feasible as the number of tumour sequencing studies increases, and as more datasets become available from matched tumour samples before and after ICB therapy. The variant database provided here will improve the interpretation of such data by enabling functional annotation of clinical variants.
IFNg signalling through the JAK-STAT pathway is not only relevant for cancer immunotherapy, but also underpins pathology in myeloproliferative neoplasms, chronic mucocutaneous candidiasis, primary immunodeficiency and several inflammatory diseases 1,2 . The molecular understanding of JAK-STAT signalling to date has been hindered by the lack of a full-length crystal structure of JAK1, and the complex intra-molecular regulation by the JAK1 pseudokinase domain 1 . We report base editing screens mapping LOF and GOF variants in key regulatory regions of the JAK1 pseudokinase-kinase domain interface, and conformational inter-molecular protein-protein interactions with SOCS1, demonstrating that base editing may be harnessed to understand complex protein biology, and potentially direct drug discovery efforts without prior detailed structural information.
Most of the functional variants discovered through base editing had clinical precedence (Supplementary Table 5), implying that immunoediting in cancer may be more prevalent than previously thought 3 . It is evident from this study and SGE experiments that mutation of key residues to any alternative residue can be deleterious 15

Cell lines and culture
All cell lines were mycoplasma tested and verified by STR profiling. Cells were maintained in a 5 % CO2, 95 % air, humidified incubator at 37 °C, in RPMI supplemented with 1X GlutaMAX, 1X penicillin-streptomycin and 10 % FCS (Thermo Fisher Scientific).
Where indicated, CellTiter-Glo proliferation assays (Promega) were performed to assess drug response following manufacturer's instructions. by Gibson assembly (NEB).

Molecular biology and cloning
All plasmid inserts were fully sequence verified by Sanger sequencing (Eurofins).
Plasmids from this article will be available from Addgene following publication (Supplementary Table 6).

Base editor cell line generation
We

Base editing screens
Base editing screens were performed with a gRNA coverage of 400-1000-fold. We for 48 h before flow cytometry analysis.

Next generation sequencing
Amplicon sequencing was performed as described 68 with primers provided in Supplementary

Validation experiments
Individual gRNAs were cloned in an arrayed format using a Golden Gate-based approach. We designed primers encoding a gRNA with BbsI overhangs and an additional G

Data analysis
To call SNPs from amplicon sequencing, we used CaVEMan 69

Giemsa staining
After six days of selection with IFNg (1500 U/ml; Thermo Fisher Scientific), cells were washed with PBS, fixed with 4 % PFA for 20 min and then stained with Giemsa working solution (1X in water; Sigma-Aldrich) for 2 h at room temperature with gentle rocking. Wells were rinsed with deionised water three times and then allowed to dry before images were taken by scanning.

Co-cultures with autologous T cells
Derivation of tumour organoids, enrichment of tumour reactive T cell populations from patient PBMCs and co-culture killing assays were performed as described 59  exonic regions. f) gRNAs targeting JAK1 exons or generating stop codons in essential genes were assigned a Rule Set 2 Score and grouped into <0.5 or >0.5. Proliferation screen zscores were compared between groups using an unpaired, two-tailed Student's ttest. g) gRNAs targeting JAK1 exons were grouped by the predicted edited cytosine's direct genomic context; preceded by a G or preceded by a T. Proliferation screen z-scores were compared between groups using an unpaired, two-tailed Student's t-test. Figure 3. Base editing mutagenesis of the IFNg pathway a) FACS gating strategy for cells with LOF in the IFNg pathway. HT-29 iBE3 cells were stimulated with IFNg (400 U/ml) for 48 h before FACS. Single cells expressing base editor (mApple) and gRNA (BFP) were gated and the cells unable to induce PD-L1 and MHC-I were gated based on a unstimulated control population. Data are representative of two independent experiments performed on separate days. b) Replicate correlation for base editor screening of the IFNg pathway. Correlation between z-scores for independent base editor screening replicate experiments performed on separate days, and independent screening assays (FACS and proliferation).

Supplementary Figure 4. Base editing reveals JAK1 LOF and GOF variants with clinical precedence a)
Replicate correlation of base editing screens using different base editor architectures and deaminases. Dot plots of gRNAs targeting JAK1 are coloured by predicted consequence. Shape indicates PAM usage of the gRNA and adjusted R 2 values are indicated. z-scores (control vs IFNg-arms; proliferation screens) are from two independent screens performed on separate days. b) Boxplot of proliferation screen z-scores for gRNAs by predicted consequence. Zscores for predicted splice variant and non-targeting gRNAs (control vs IFNg-arms) were compared using an unpaired, two-tailed Student's t-test. Shown is the median, box limits are upper and lower quartiles, whiskers are 1.5× interquartile range, and points are outliers. c) Heatmap amino acid substitution matrix, showing aggregated predicted codon changes for each gRNA targeting JAK1 and gRNA z-scores from control vs IFNg-arms for BE3.9max-NGN and ABE8e-NGN proliferation screens. d) Comparison of bioinformatic prediction of variant effect with experimental data from base editing screens (z-scores from control vs IFNg-arms; proliferation screens). SIFT (0 is deleterious, 1 is tolerated), PolyPhen (0 is benign, 1 is damaging) and BLOSUM62 (positive is conserved, negative is not conserved).
Supplementary Figure 6. Amplicon sequencing of JAK1 following base editing Amplicon sequencing of endogenous JAK1 DNA reveals the editing profile of BE3 gRNAs. Position of edits relative to the protospacer are shown for LOF and GOF gRNAs in the validation cohort. Data are generated from control cells, cells with base editing or base editing and selection with IFNg for 6 d. Data represent the mean of two independent experiments performed on separate days.

CRISPR-Cas9 screening identified druggable targets that sensitised tumour cells to
IFNg when inactivated, such as MCL1 and TBK1, highlighting potential ICB-combination therapies in CRC. In line with this, TBK1 inhibition has been reported to increase immune reactivity to tumour organoids ex vivo 1 . Conversely, we revealed that mTOR inactivation can facilitate tumour-intrinsic resistance to IFNg, arguing against combining mTOR inhibitors with ICB in CRC 2 , although our reductionist approach does not consider the potential effects of these drugs on immune cells. Interestingly, inactivation of KEAP1, FBXW7, NF2, and STK11, modulated sensitivity to IFNg, emphasising important non-cell autonomous roles for these tumour suppressor genes. KO of NF2 resulted in increased resistance to IFNg, and has also been linked to BRAF inhibitor resistance 3,4 , consistent with an overlap between ICB resistance and MAPK inhibitor resistance pathways 5 , with possible implications for the efficacy of ICB in melanoma patients pre-treated with BRAF inhibitors.