In this talk, Pavel Tolar (Francis Crick Institute) outlines the rationale, methodology, and outcomes of genome‑wide pooled CRISPR‑Cas9 screens applied to B lymphocytes. He begins with a concise overview of B‑cell biology, emphasizing how antigen‑driven signaling networks dictate fate decisions (activation, proliferation, plasma‑cell differentiation) and how dysregulation leads to immunodeficiency, autoimmunity, allergy, or malignancy. He then critiques legacy genetic tools (siRNA, TALENs) as insufficient for high‑throughput interrogation of immune cells, setting the stage for CRISPR‑Cas9 as a robust, scalable solution. The core of the presentation details the pooled screening workflow: library construction (4–10 guides per gene, 50–200 k guides), viral transduction to achieve one guide per cell, selection (drug markers or fluorescence), assay readouts (growth, negative‑selection, FACS‑based phenotyping), and sequencing‑based quantification of guide abundance. He explains the computation of CRISPR scores (and a modified internalization score) to infer gene effects, and the statistical pipeline for hit calling. Two concrete screens are showcased: (1) an antigen‑internalization screen in Ramos B cells that identified ~80 regulators, highlighted by SH3GL1 (endophilin A2) mediating a clathrin‑independent endocytic pathway; in vivo validation in endophilin‑deficient mice demonstrated impaired antigen uptake, reduced B‑cell numbers, and diminished IgG responses. (2) a class‑switching screen comparing IgE‑ versus IgG1‑expressing B cells, which uncovered mTOR as a positive regulator of IgE plasma‑cell differentiation and, unexpectedly, calcium‑calcineurin signaling (via PPP3R1) as a negative regulator. Targeted knockout of PPP3R1 in hematopoietic stem cells increased IgE‑plasma cells without affecting IgG1, confirming the inhibitory role of calcium signaling in IgE responses. Tolar concludes that pooled CRISPR screens are the most efficient platform for genome‑wide functional studies in primary immune cells, and he outlines future extensions: single‑cell readouts, dual‑targeting for interaction mapping, CRISPRi/a for transcriptional modulation, and community data sharing for meta‑analyses. Acknowledgments to lab members and collaborators close the talk.
Chapters
Hello everybody, thank you for connecting to this virtual session and to listening to my talk. I am Paweł Tolar, I'm a group leader at the Francis Crick Institute in London, and I'm going to talk today about the principles of genome-wide CRISPR-Cas9 screening with particular focus on immune cells. I will start by introducing you to the role of B lymphocytes in antibody responses, the topic of research of my lab. I explain what the previous limitations were in genetic manipulation in immune cells, then introduce the principles of the new pooled CRISPR Cas9 screening, and then go into detail of the design, execution, and analysis of these screens. I will then give you examples of the screens that we have done in the lab for various functions in B lymphocytes, and I will close with conclusions and outlook for this technology into the future. Okay, so my lab is interested in how immune cells detect and respond to pathogens, and we are particularly focusing on the antibody response. As you know, antibody production is initiated by B lymphocytes or B cells, and B cells are present in large numbers in lymphoid organs, and each of them carries a unique B cell receptor on this surface that carries a different specificity for antigen. So the presence of a certain pathogen pathogen selects and activates certain B cells, certain B cell clones from the naive repertoire, and these B cell clones expand, they proliferate, and eventually differentiate into antibody-secreting cells called plasma cells that produce large amounts of antibody that can fight off the pathogen. So as you can see, this response is mediated by decisions of individual B cells to activate, to divide, to differentiate into plasma cells, and eventually to die. Now, on the molecular level, these decisions are mediated by a complex intracellular signaling network. The network is triggered by binding of foreign antigens to the B cell receptor and then involves multiple interconnected signaling branches which also take input from other inflammatory signals. The decisions also require B cell interacting with helper T cells, and that requires another function of the B cell receptor to physically capture and endocytose the antigens and deliver them into antigen processing compartments where antigenic peptides can be loaded onto MHC class II proteins. And the display of MHC class II proteins on B cell surfaces allows collaboration and interaction with these helper T cells. So the complexity of the network reflects the complexity of the immune system and the challenge to make the fate decisions correctly, and we know that even a slight imbalance in these networks can turn protection from pathogens into immune disease. For example, defects that reduce the ability of B cells to respond will leave the system in immune deficiency, Because the B cell repertoire is generated randomly, there's always cells that are self-reactive, and these have to be suppressed correctly, otherwise they can cause autoimmune diseases. B cells can also respond to material that is foreign but otherwise innocuous, such as allergens, and this could lead to overproduction of IgE and allergic disease. And because the generation of the B cell repertoire involves DNA recombination and somatic mutations, inappropriately activated B cells are also a common source of B cell malignancies in the form of leukemias and lymphomas. So to understand how B cells respond to antigens, we need to understand how the signaling decision networks inside of them work, and we now know this involves thousands of genes, so we would really benefit from high-throughput robust genetic approaches. Unfortunately, the previous technologies were not up to the task. The screening that had been developed in lower organisms cannot be applied to the mammalian immune system, and manipulation of somatic cells through approaches such as siRNA or TALENs has turned out to be quite inefficient in immune cells and B cells particularly. So they were again quite inadequate for high-throughput screening approaches. So this all changed with the discovery of the CRISPR-Cas9 DNA targeting system, which I'm sure you're very familiar with. The advantage of the system is that it does not rely on any endogenous machinery, but brings directly the Cas9 nuclease via a guide RNA to a site, specific site, on the DNA, and the advantage is that the system is very efficient and works very robustly even in immune cells, and this opened the field to high-throughput screening approaches. There are two screening designs that can be used with CRISPR/Cas9 approaches. The first is arrayed library screening, in which essentially each is read individually in a well in a plate, and that allows screening for a broad range of phenotypes including time-lapse and dynamic assays, but it's expensive and requires high-throughput sample handling facilities as well as extensive controls and normalization. But because of the presence of the guides in the cells can be read out by sequencing, this technology is also amenable to pooled library screening. The pooled library screening is much less expensive and does not require any specialized equipment. It's limited to one readout per screen and also can screen only for cell intrinsic functions, but it is by far the simplest genome-wide screening technology available today. So how does pooled library screening work? Well, it starts with a library of plasmids, each containing one guide RNA that is typically packaged into viruses to be able to transduce them into cells in such a way that results in each cell containing one guide RNA. The cells are then selected for the presence of the guides, and they are subjected to an assay, which can be just growth and survival assay, or it could be some sort of assay that's able to isolate or sort out responders from non-responders. And the different cell populations are then sequenced for the presence of the guides, And from the frequency of the guide RNAs, you can then measure the effect of the gene, of the corresponding gene, on that particular cellular process. So let's look at these individual steps in a little bit more detail. The first thing to consider is the scale of the experiments. Whole genome libraries typically contain 4 to 10 guide RNAs per gene, so that results in 50 to 200,000 plasmids or guide RNAs in the library. So that means that both the plasmids and the cells need to be handled at scale to maintain the complexity in the library. So in cell numbers, that translates to carrying out the experiments with approximately 10 to 200 million of cells. The complexity of the library is quite easily assayed again by high-throughput sequencing of the guide RNAs, where you're looking for this normal distribution of the guide RNA frequencies with very little or very few guides missing from the library. The second thing to consider is how are cells containing guide RNAs selected and how the guide targeting efficiency is assayed. So selection is typically done using drug markers present in the plasmid, but also can be done using fluorescent sorting. The targeting efficiency can be assessed by analyzing the frequency of guide RNAs that target known essential genes. So for this, you can compare the guide RNA frequency in the cell population to that in the original library using this formula shown here and calculate so-called CRISPR score.. So if you then plot the CRISPR score for various genes in the library, you can see what the genes, what the effect of the genes on the cells have been. So for non-targeting controls, the values cluster around 0, as these guides are having no effect on the cells, but as shown here in blue for top 200 known essential genes, the CRISPR scores are well below zero, and that is a measure of the targeting efficiency in the experiment. Perhaps the most important step in the screen is design of the readout. There are three types of experiments. In the first one, the cells are essentially assayed for growth and so can identify genes that are essential. In the second one, you can apply negative selection during the growth resulting in overgrowth only of a few select clones. And so these screens can identify genes important for sensitivity to that treatment. But you can also subject the cells to assays that allow you eventually to sort the cells into two populations, for example, using flow cytometry. And this way you can screen for almost any cell-intrinsic phenotype. The final consideration is for the analysis of the screen results. As I explained before, the screens are read out by sequencing of the guide RNAs and by comparing their frequencies in the different populations, and this is expressed by this CRISPR score as shown here. The CRISPR scores of individual guide RNAs are then averaged per gene, and that gives you the strength of the effect of each gene on the assay phenotype. And the statistical significance of that gene effect is typically derived from ranking the individual guide RNA CRISPR scores against all the other CRISPR scores in the screen. So from all these results, you can eventually call out statistically significant hits. In the screen. Okay, now I want to show you two examples of screens that we've done in B lymphocytes just to give you a better flavor for the implementation of these screens and the results you can expect. The first screen was focused on endocytosis of the B cell receptor, which I explained is a critical function in B cells that establishes collaboration with helper T cells.. But previously the composition, the molecular composition of the endocytic pathways in B cells was a big unknown. So for this screen, we used a pooled guide RNA library that was delivered by a lentiviral transduction into a B cell line called Ramos that already expressed Cas9 protein. We then subjected these Ramos B cells to a fluorescent antigen internalization assay that allowed us to sort cells that internalized very little of the antigen or a lot. And by comparing these by DNA sequencing, we can then quantify the effect of each gene on internalization by using this modified CRISPR score that we called internalization score. This is how the results looked like when we plotted the internalization scores against the the statistical significance of the genes. We used two different libraries in these screens, and while there was a significant overlap between the results, there was also a number of differences. So to be able to validate as many genes as possible, we decided to take top 200 genes from either of these screens and design a smaller validation library.. The increased statistical power of this smaller library allowed us to robustly identify about 80 genes that were implicated in regulation of antigen uptake in these cells. And this is the first comprehensive mapping of the endocytic and intracellular trafficking pathways in B cells downstream of the B cell receptor. Quite surprisingly, the genes identified cover quite a wide range of predicted biological functions as shown in the cell model on the right. But I want to focus today on just one of them to illustrate their importance for the immune response, and the genes that— the gene that we focused on was called SH3GL1. SH3GL1 encodes a protein called endophyllin A2, which has recently been implicated in mediating clathrin-independent endocytosis. And by studying this protein, we could indeed establish that the B cell receptor is using this endophyllin-mediated endocytosis in addition to the previously known clathrin-mediated endocytosis. And just to illustrate the importance of this novel endophyllin A2 mediated endocytic pathway for B cell responses, I want to show you a couple of results from endophyllin-deficient mice. So we first validated that the B cells from endophyllin-deficient mice have slower uptake of antigen. By using GFP-labeled endophyllin, we could also establish that the endophyllin forms endocytic spots on B cell membrane colocalizing with antigen but forming independently of clatrin. The spleens of these endophyllin-deficient mice were smaller because they had less B cells, and these B cells were poor to respond to antigen stimulation. So after immunization, there were lower numbers of activated antigen-specific B cells, and the production of antigen-specific IgG was decreased. So the screen identified a novel claptin-independent endocytic pathway downstream of the B cell receptor that is essential for B cell responses to antigen, particularly at a step that requires interaction with T cells. And for the future, we are interested in understanding exactly how endophyllin-8 works in B cells and whether, in addition to the B cell receptor mediates trafficking of other cargoes that may be important for B cell responses. The second screen I want to show you is focused on the signaling downstream of the receptor that regulates the fate decisions that B cells make during responses. And in particular, it's focused on the fact that as B cells are activated, they switch their membrane immunoglobulin of the BCR from IgM to other isotypes such as IgG, IgA, and IgE. And it's known that the identity of this membrane immunoglobulin alters the signaling networks downstream of the BCR, and it influences the fate decisions that these class-switched B cells make. But it's not clear exactly how these differences are encoded in the signaling networks. So we focused on IgE the least abundant but also least understood of these isotypes. To screen for genes that regulate IgE B cell responses, we used a modified guide RNA library that contains mCherry as a marker, and we transduced this library into primary B cells isolated from mice that express Cas9 protein labeled with GFP. B. And so when these cells are cultured in a feeder-based culture that mimics T cell help, these B cells switch to express on their surface either IgG1 B cell receptor or IgE B cell receptor, and each of these B cells can be induced to differentiate into plasma cells. So by sorting these 4 populations and comparing them by CRISPR score calculation, we could infer the importance of individual genes on the IgE B cells compared to their IgG1 counterparts, and also on the process of plasma cell differentiation. This slide shows the results by plotting the CRISPR scores for the individual populations focusing on the IgE B cells. On the left is a comparison between IgE B cells and IgG1 B cells, and this analysis identified some common essential genes required in both of these populations shown in red, and these are expected from the immunology in the system. But it also identified genes that were selectively involved in IgE cells, either suppressive or essential, shown in orange and green, and these sets of genes mostly were centered on apoptotic pathways. We could also analyze differentiation of the IgE B cells into plasma cells, showed on the right, and this analysis again identified some known regulators of the process but also identified a large number of genes that were unexpected and that we analyzed further. Through pathway analysis and validation, we could link these genes into two pathways shown here in this slide. So on the left side, mostly in blue, is a positive regulatory pathway mediated by mTOR signaling, and on the right is a negative regulation of IgE plasma cell differentiation by calcium signaling. And the negative role of calcium was really surprising as previously calcium was thought to be a universally positive signal for B cell activation, but our results predicted that calcium signaling, particularly through calcineurin, is actually inhibitory for IgE B cells. To validate the results in vivo, we designed a system for rapid gene targeting in mouse hematopoietic stem cells, which upon transplantation can be then traced through immune cell development and responses to immunization. And the system works very well, it's very specific and efficient. We use CD4 targeting as a control, which does not affect the numbers of B cells but does abrogate the presence of CD4-positive T cells.. And then here we are targeting the essential regulatory subunit of calcineurin encoded by PPP3R1. So you can see that also has an effect on T cells, and that's expected from the literature. But the critical result is the targeting of this calcineurin subunit increases the number of IgE-positive plasma cells upon immunization while not changing changing the numbers of IgG1 plasma cells. So indeed calcium is an inhibitory pathway for IgE responses. In summary, this CRISPR screen provides the most comprehensive identification of genes essential for class-switched B cell responses to date. It allowed us to do a differential analysis to identify genes selectively involved in IgG1 or IgE B cells. And together with this in vivo validation pipeline, I think can provide us new ideas how to target these individual class-switched B cell responses selectively. And this can be interesting, for example, in selective dampening of IgE responses in allergy. This is all I wanted to show you and brings me to my conclusions. The pooled CRISPR-Cas9 screens are currently the most efficient way to carry out genome-wide functional studies, and I think the results are so efficient that they go beyond just identification of rare hits in those screens, but they really can be used to measure gene effects on a variety of cellular functions. The screens are directly applicable to primary cells and can be carried out in disease contexts. And there's some exciting ways how to extend them in the future just beyond improvements in the design and statistical analysis. For example, there are now designs to carry out these screens using single-cell sequencing for transcriptional readout in the individual cells and for mapping of transcriptional circuits. The screens are also compatible with dual targeting, which allows high-throughput mapping of gene interactions. And they're also compatible with modified Cas9 proteins, for example, to carry out CRISPRi or CRISPRa type of gene targeting or transcriptional modulation. And finally, I think that public sharing of the screen data really provides another level of analysis that cannot be carried out in a single lab and allows meta-analysis that can identify context-specific genes in a really efficient way. Finally, I want to acknowledge the people in the lab who did the experiments I showed you today. Despite the relative simplicity, the genome-wide screens are still a huge amount of work. Destiny Malinova, former postdoc in the lab, carried out the endocytic screens and all the in vivo follow-up, and she was helped by Labia Wasim, who did the endophyllin imaging. And Becky Newman is a postdoc in the lab who is carrying out the IgG1 and IgE screens with the in vivo follow-up. I also want to thank our funders and our collaborators and you for your attention. Thank you very much.
