Supplemental Figure 1. IRF4-GFP knock-in reporter cell lines. Supplemental Figure 2. IRF4-GFP CRISPR screen in adult T-cell lymphoma and screen controls. Supplemental Figure 3. The OST is essential for DLBCL. Supplemental Figure 4. The role of the OST complex in radiation chimeras. Supplemental Figure 5. The B-cell receptor is glycosylated by the OST complex. Supplemental Figure 6. B-cell receptor deglycosylation leads to higher surface expression. Supplemental Figure 7. B-cell receptor deglycosylation leads to signaling changes in proximal BCR signaling components. Supplemental Figure 8. BCR deglycosylation leads to recruitment of CD22 on the cell surface to the BCR and signaling alterations in GCB DLBCL. Supplemental Figure 9. BCR internalization and My-T-BCR formation in glycosylation deficient cells. Supplemental Figure 10. NGI-1 synergizes with BTK and PI3K inhibitors.

Table S1. Brunello library sgRNA normalized read counts of IRF4 sorted CRISPR screens. Table S2. Segregation score of IRF4 sorted CRISPR screens. Table S3. Glycoproteomics in STT3A or STT3B KO cells. Table S4. Brunello sgRNA Library Normalized Read Counts of CRISPR screens. Table S5. CRISPR screen score of essentiality screens. Table S6. Glycoproteomics after NGI-1 treatment. Table S7.HBL1 and TMD8 Phosphotyrosone Phosphoproteome (pYome). Table S8. TMD8 CD79A-BioID. Table S9. HBL1 CD22 BioID. Table S10. Brunello library sgRNA normalized read counts of NGI-1 drug modified screen (TMD8). Table S11. CRISPR screen score of NGI-1 drug modifier screen. Table S12. TMD8 Drug screen Excess HSA.

Diffuse large B-cell lymphomas (DLBCL) are genetically and clinically heterogenous. Despite advances in the gene-expression and genomic characterization, the pathophysiology of high-risk DLBCL still remains incompletely understood. We report the first proteogenomic analysis of DLBCL tissues from 438 patients to elucidate disease pathophysiology, delineate so far undescribed high-risk tumor characteristics and inform diagnostic and therapeutic approaches. By integration of genetic, transcriptomic and mass-spectrometry-based quantitative proteomic data using machine learning approaches and latent variable modelling, we identify seven DLBCL proteogenotypes (PG) reflecting specific pathophysiological patterns and spanning cell-of-origin (COO) boundaries. These PGs differed regarding their cellular and immune microenvironment as well as B-cell-intrinsic oncogenic programs and differentiation patterns and could not be identified by mere genomics or transcriptomics. PG4 was discovered as high-risk DLBCL with inferior overall and progression-free survival upon R-CHOP-based therapy independently of known risk factors such as COO, genetic subtypes, lymphoma ecotypes and the international prognostic index. PG4 was validated as an independent risk factor in additional cohorts of de novo DLBCL (n=865 patients). It was enriched for specific subsets of activated B-cell-like DLBCL with MYD88/CD79b mutations (68.4 % of all ABC-MCD/MYD88 DLBCLs) and double-hit-signature-positive germinal center B-cell-like tumors (18.6 % of all Dhit-sig+ GCB DLBCLs) indicating heterogeneity within these genetic subtypes. PG4 cases shared a dark-zone related B-cell phenotype, an enrichment of BTG1 and TBL1XR1 mutations and enhanced B-cell receptor and MYC activity. Moreover, deregulated protein translation was a PG4-defining feature evident at the proteomic level only. Single-cell RNA and ATAC sequencing of 43 DLBCL cases (197,860 cells) with proteogenotype information confirmed enhanced BCR, MYD88 and PI3K activity and the dark-zone phenotype in the PG4 malignant B cells. Moreover, TCF3 and TCF4 transcription factors showed enhanced activity in GCB-PG4 and ABC-PG4 tumors, respectively. Analysis of the PG4 tumor microenvironment at single-cell granularity revealed a significant T-cell depletion and exhaustion of the CD8 T-cell effector memory compartment, likely contributing to resistance against conventional chemoimmunotherapy. Importantly, PG4 was detectable even at the subclonal level in single-cell sequencing data from individual tumors (n=103 patients, 504,444 cells). Here, 24 tumors had PG4-predicted subclones that biologically varied significantly from other subclones in the tumors (p<0.01), providing for the first time evidence for intratumoral proteogenomic heterogeneity. In addition to identifying PG4-DLBCLs as chemo-resistant tumors, we provide insight into how the identified PGs respond to innovative immunotherapies such as CAR T-cell therapy. In summary, our study provides an integrated proteogenomic framework up to the single cell level explaining so far unresolved disease heterogeneity and identifying molecular features of high-risk DLBCL as a basis for innovative diagnostic and therapeutic approaches. The diagnostic detection of PG4-DLBCL as chemo-resistant lymphoma has the potential to improve patient outcome by enabling the selection of more efficient therapies for these tumors.

Glucocorticoids have been used for decades to treat lymphomas without an established mechanism of action. Using functional genomic, proteomic, and chemical screens, we discover that glucocorticoids inhibit oncogenic signaling by the B cell receptor (BCR), a recurrent feature of aggressive B cell malignancies, including diffuse large B cell lymphoma and Burkitt lymphoma. Glucocorticoids induce the glucocorticoid receptor (GR) to directly transactivate genes encoding negative regulators of BCR stability (LAPTM5; KLHL14) and the PI3 kinase pathway (INPP5D; DDIT4). GR directly represses transcription of CSK, a kinase that limits the activity of BCR-proximal Src-family kinases. CSK inhibition attenuates the constitutive BCR signaling of lymphomas by hyperactivating Src-family kinases, triggering their ubiquitination and degradation. With the knowledge that glucocorticoids disable oncogenic BCR signaling, they can now be deployed rationally to treat BCR-dependent aggressive lymphomas and used to construct mechanistically sound combination regimens with inhibitors of BTK, PI3 kinase, BCL2, and CSK.

Prednisone is an anti-inflammatory glucocorticoid (GC) that is cytotoxic for normal and malignant B cells and, on this basis, has long been included in combination chemotherapies to treat aggressive B-cell lymphomas, including diffuse large B cell lymphoma (DLBCL) and Burkitt lymphoma (BL). GCs act by binding to the glucocorticoid receptor (GR; NR3C1), a ligand-induced transcription factor. The transcriptional response to GCs can vary significantly due to the variation in expression levels of GR cofactors and chromatin landscapes in different cell types. The mechanisms by which GCs kill malignant lymphoma cells are largely unknown, prompting us to search for new targets of GC action with the hypothesis that GCs may inhibit key survival pathways in lymphomas. To identify genes that synergize or antagonize GC lethality in lymphomas, we performed genome-wide CRISPR-Cas9 screens in the presence and absence of prednisolone, the active metabolite of prednisone. Screens in cell line models of BL and DLBCL (both ABC and GCB subtypes) revealed strong synergy between GC treatment and inactivation of genes encoding components of the B cell receptor signaling pathway, which is required to sustain the viability of these malignant lymphoma cells. With combination of the cleavage under targets and release using nuclease (CUT&RUN) assay and RNA-seq, we identified GR binding to the LAPTM5 locus at its glucocorticoid response element and AP1 motifs upon GC treatment. GR binding also induced expression of LAPTM5, which negatively regulated BCR signaling by promoting the lysosomal degradation of the BCR. Conversely, GC induced binding of GR to the CSK locus, thereby repressing expression of the non-receptor tyrosine kinase CSK, which antagonizes BCR signaling by phosphorylating an inhibitory tyrosine residue present in all Src-family kinases (SFKs). However, in BCR-dependent aggressive lymphomas, inactivation of CSK paradoxically decreased proximal BCR signaling and induced cell death. By performing quantitative phosphoproteome and ubiquitinome with mass spectrometry, we demonstrated that treatment of lymphoma models with a small molecule inhibitor of CSK kinase activity (CSKi) initially increased constitutive BCR signaling, as expected, but then triggered exuberant ubiquitination of the LYN, HCK and BLK, leading to their proteasomal degradation. Consequently, the CSKi blocked BCR-dependent NF-kB activation in ABC DLBCL models and BCR-dependent PI3 kinase activation in models of GCB DLBCL and BL. In summary, inhibition of oncogenic BCR signaling is a major mode of action for GCs, which have been used empirically for decades to treat lymphomas. GCs restrain the most proximal steps in BCR signaling at the plasma membrane by, on one hand, decreasing BCR abundance, and on the other hand, by decreasing CSK expression, thereby reducing expression of the essential SFKs. Small molecule inhibition of CSK kinase activity potentiated the effect of GCs on oncogenic BCR signaling and strongly synergized with GCs in killing ABC and GCB DLBCL models in vitro and preventing the growth of ABC and GCB DLBCL and patient-derived xenografts, warranting the development of clinical-grade CSK inhibitors for the treatment of these aggressive cancers.

AbstractDiffuse large B-cell lymphoma (DLBCL) can be subdivided into the activated B-cell (ABC) and germinal center B cell–like (GCB) subtypes. Self-antigen engagement of B-cell receptors (BCR) in ABC tumors induces their clustering, thereby initiating chronic active signaling and activation of NF-κB and PI3 kinase. Constitutive BCR signaling is essential in some GCB tumors but primarily activates PI3 kinase. We devised genome-wide CRISPR–Cas9 screens to identify regulators of IRF4, a direct transcriptional target of NF-κB and an indicator of proximal BCR signaling in ABC DLBCL. Unexpectedly, inactivation of N-linked protein glycosylation by the oligosaccharyltransferase-B (OST-B) complex reduced IRF4 expression. OST-B inhibition of BCR glycosylation reduced BCR clustering and internalization while promoting its association with CD22, which attenuated PI3 kinase and NF-κB activation. By directly interfering with proximal BCR signaling, OST-B inactivation killed models of ABC and GCB DLBCL, supporting the development of selective OST-B inhibitors for the treatment of these aggressive cancers.Significance:DLBCL depends on constitutive BCR activation and signaling. There are currently no therapeutics that target the BCR directly and attenuate its pathologic signaling. Here, we unraveled a therapeutically exploitable, OST-B–dependent glycosylation pathway that drives BCR organization and proximal BCR signaling.This article is highlighted in the In This Issue feature, p. 1749

Tumors frequently subvert major histocompatibility complex class I (MHC-I) peptide presentation to evade CD8+ T cell immunosurveillance, though how this is accomplished is not always well defined. To identify the global regulatory networks controlling antigen presentation, we employed genome-wide screening in human diffuse large B cell lymphomas (DLBCLs). This approach revealed dozens of genes that positively and negatively modulate MHC-I cell surface expression. Validated genes clustered in multiple pathways including cytokine signaling, mRNA processing, endosomal trafficking, and protein metabolism. Genes can exhibit lymphoma subtype- or tumor-specific MHC-I regulation, and a majority of primary DLBCL tumors displayed genetic alterations in multiple regulators. We established SUGT1 as a major positive regulator of both MHC-I and MHC-II cell surface expression. Further, pharmacological inhibition of two negative regulators of antigen presentation, EZH2 and thymidylate synthase, enhanced DLBCL MHC-I presentation. These and other genes represent potential targets for manipulating MHC-I immunosurveillance in cancers, infectious diseases, and autoimmunity.

Rationale: Airways obstruction with thick, adherent mucus is a pathophysiologic and clinical feature of muco-obstructive respiratory diseases, including chronic obstructive pulmonary disease, asthma, and cystic fibrosis (CF). Mucins, the dominant biopolymer in mucus, organize into complex polymeric networks via the formation of covalent disulfide bonds, which govern the viscoelastic properties of the mucus gel. For decades, inhaled N-acetylcysteine (NAC) has been used as a mucolytic to reduce mucin disulfide bonds with little, if any, therapeutic effects. Improvement of mucolytic therapy requires the identification of NAC deficiencies and the development of compounds that overcome them. Objectives: Elucidate the pharmacological limitations of NAC and test a novel mucin-reducing agent, P3001, in preclinical settings. Methods: The study used biochemical (e.g., Western blotting, mass spectrometry) and biophysical assays (e.g., microrheology/macrorheology, spinnability, mucus velocity measurements) to test compound efficacy and toxicity in in vitro and in vivo models and patient sputa. Measurements and Main Results: Dithiothreitol and P3001 were directly compared with NAC in vitro and both exhibited superior reducing activities. In vivo, P3001 significantly decreased lung mucus burden in beta ENaC-overexpressing mice, whereas NAC did not (n = 6-24 mice per group). In NAC-treated CF subjects (n = 5), aerosolized NAC was rapidly cleared from the lungs and did not alter sputum biophysical properties. In contrast, P3001 acted faster and at lower concentrations than did NAC, and it was more effective than DNase in CF sputum ex vivo. Conclusions: These results suggest that reducing the viscoelasticity of airway mucus is an achievable therapeutic goal with P3001 class mucolytic agents.

Bacterial chemosensory signal transduction systems that regulate motility by type IV pili (T4P) can be markedly more complex than related flagellum-based chemotaxis systems. In T4P-based systems, the CheA kinase often contains numerous potential sites of phosphorylation, but the signaling mechanisms of these systems are unknown. In Pseudomonas aeruginosa, the Pil-Chp system regulates T4P-mediated twitching motility and cAMP levels, both of which play roles in pathogenesis. The Pil-Chp histidine kinase (ChpA) has eight Xpt domains; six are canonical histidine-containing phosphotransfer (Hpt) domains and two have a threonine (Tpt) or serine (Spt) in place of the histidine. Additionally, there are two stand-alone receiver domains (PilG and PilH) and a ChpA C-terminal receiver domain (ChpArec). Here, we demonstrate that the ChpA Xpts are functionally divided into three categories as follows: (i) those phosphorylated with ATP (Hpt4-6); (ii) those reversibly phosphorylated by ChpArec (Hpt2-6), and (iii) those with no detectable phosphorylation (Hpt1, Spt, and Tpt). There was rapid phosphotransfer from Hpt2-6 to ChpArec and from Hpt3 to PilH, whereas transfer to PilG was slower. ChpArec also had a rapid rate of autodephosphorylation. The biochemical results together with in vivo cAMP and twitching phenotypes of key ChpA phosphorylation site point mutants supported a scheme whereby ChpArec functions both as a phosphate sink and a phosphotransfer element linking Hpt4-6 to Hpt2-3. Hpt2 and Hpt3 are likely the dominant sources of phosphoryl groups for PilG and PilH, respectively. The data are synthesized in a signaling circuit that contains fundamental features of two-component phosphorelays.