Improving CRISPR-Cas9 Screens in CAR T Cells: A Refined Method for Library Preparation

Summary

We developed an optimized CRISPR-Cas9 knockout screening protocol for primary CAR T cells. By reducing gDNA carryover through enzymatic digestion and sgRNA cassette pulldown, this approach minimizes PCR artifacts, ensuring accurate sgRNA detection and robust identification of functional gene targets.

Abstract

Chimeric antigen receptor (CAR) T cell therapies have demonstrated remarkable efficacy in several hematological malignancies, yet their success has not been fully replicated in solid tumors. Moreover, even in hematological cancers, relapse after CAR T cell infusion continues to compromise long-term outcomes. These challenges highlight the urgent need to develop strategies that enhance CAR T cell efficacy, persistence, overcoming tumor and microenvironment-mediated resistance. Clustered regularly interspaced short palindromic repeats (CRISPR)-Cas9-based screening platforms provide a powerful approach to systematically identify genes that regulate CAR T cell function. By linking genetic perturbations to phenotypic outcomes, these assays enable the discovery of pathways controlling activation, proliferation, memory formation, and cytotoxicity. Standard workflows involve transduction of substantial numbers of cells with a single guide RNA (sgRNA) library, Cas9-mediated editing, selection of edited cells, and PCR amplification of sgRNA cassettes from genomic DNA (gDNA) prior to sequencing. However, PCR amplification using large amounts of gDNA poses significant challenges and often fails to selectively amplify and retrieve sgRNAs. Here, we describe an optimized CRISPR-Cas9 knockout screening protocol, which we have tested on primary human CAR T cells. The method here incorporates an intermediate step during sgRNA library preparation that reduces gDNA carryover through enzymatic digestion and selective pulldown of the sgRNA cassette, thereby increasing the efficiency of the first PCR amplification. This modification allowed us to retrieve sgRNA information across our CAR T cell screens, which had remained elusive in our previous attempts using traditional 1 and 2-step PCR amplification protocols. In conclusion, this optimized workflow facilitates CRISPR screening library preparation in challenging samples and enables the identification of key genetic determinants that can be targeted to improve therapeutic efficacy.

Introduction

Immunotherapy has revolutionized cancer treatment, offering novel strategies to harness and modulate the immune system. Among these, chimeric antigen receptor (CAR) T cell therapy has emerged as one of the most transformative approaches. This therapy involves the genetic engineering of a patient's T cells ex vivo with a synthetic receptor designed to recognize a specific tumor antigen, thereby enhancing antitumor activity¹. CAR T cells have shown remarkable success in hematological malignancies, with outcomes ranging from short-lived remissions to prolonged disease-free survival with minimal toxicity in patients receiving CD19 or BCMA-targeted therapies². To date, the FDA has approved seven CAR T cell products for hematological cancers³. Despite these advances, significant limitations remain. Challenges arise from the CAR T cell product itself, often influenced by the manufacturing process or the fitness of the patient's T cells, which may be impaired by disease progression, prior therapies, or age. Tumor-intrinsic resistance mechanisms, such as antigen downregulation, can also compromise efficacy⁴. Furthermore, the immunosuppressive tumor microenvironment poses an additional barrier that severely limits CAR T cell persistence and function, particularly in patients with solid tumor⁵. Consequently, there is an urgent need to enhance both the initial and long-term efficacy of CAR T cell therapies.

The advent of CRISPR-Cas9 gene editing has provided powerful tools to dissect gene function through large-scale screening approaches. In these assays, cells are transduced with a single-guide RNA (sgRNA) library, typically delivered by lentiviral vectors, ensuring one sgRNA per cell and stable genomic integration. Following Cas9-mediated editing and selection of transduced cells, genomic DNA (gDNA) is extracted, and sgRNA cassettes are amplified by PCR for library preparation. High-throughput sequencing then enables the quantification of sgRNA distributions across different phenotypes, thereby identifying genes that positively or negatively regulate the process under study⁶.

CRISPR screening has been widely applied to explore T cell biology and, more recently, to enhance CAR T cell performance. Genome-wide screens have identified regulators of fundamental T cell processes, including activation, proliferation, and differentiation. For example, FAM49B was identified as a regulator of T cell receptor signaling⁷, while SOCS1, TCEB2, RASA2, and CBLB were shown to be essential for proliferation following stimulation⁸. Beyond these core pathways, CRISPR screening has also elucidated genes involved in T cell memory and exhaustion. In vivo studies identified Fli1 as a candidate to enhance effector responses without disrupting memory or exhaustion precursors⁹, and the chromatin remodeler Arid1a as a regulator whose loss reduces T cell exhaustion¹⁰. Regulators of T helper type 2 (Th2) differentiation have also been characterized with this approach¹¹. Using a custom sgRNA library targeting 25 kinases, p38 was found to promote expansion, memory formation, and protection from oxidative and genomic stress¹². Similarly, REGNASE-1 knockout in CD8⁺ T cells conferred a long-lived effector phenotype that improved tumor control in melanoma and leukemia models¹³. Additional genome-wide screens identified Dhx37 as a regulator of T cell activation and cytotoxicity¹⁴, while LTBR and other genes were validated as enhancers of T cell function in CAR T and γδ T cells¹⁵. More recently, CRISPR screening has been applied directly in CAR T cells, revealing novel targets such as PRODH2, an enzyme involved in proline metabolism that enhances CAR T antitumor activity¹⁶, and TLE4 and IKZF2, whose inactivation improved CAR T efficacy in glioblastoma models¹⁷.

A critical step in CRISPR screens is precisely the selective amplification of sgRNA cassettes from exceedingly large amounts of gDNA, a challenge difficult to circumvent given the vast number of cells used in CRISPR screenings. PCR amplification from such large amounts of DNA carryover poses several technical challenges, including molecular crowding that limits the physical diffusion of DNA and polymerase molecules, transient non-specific binding of the primers and the polymerase to the DNA background, off-target amplification resulting in primer depletion, or Mg²⁺ sequestration by the DNA phosphate backbone, among other challenges^18,19,20,21. These often result in PCR failure and/or bias. To mitigate these issues, different strategies have been proposed, including separating sgRNA amplification from adaptor addition into two PCR steps²², using target enrichment with biotinylated oligos and magnetic bead capture^23,24, or designing plasmids with restriction sites flanking the sgRNA cassette for fragment enrichment²⁵.

In this study, we present an optimized CRISPR-Cas9 knockout screening protocol for primary human CAR T cells. The CRISPR library used is Brunello Kinome 1, a library that contains 3052 unique 20nt-long sgRNAs targeting 763 human kinase genes (4 sgRNAs per target). T cells were isolated through CD4⁺ and CD8⁺ magnetic positive selection from the peripheral blood mononuclear cells (PBMC)-enriched fraction of blood samples from healthy donors. Upon activation with anti-CD3 and anti-CD28, T cells were spinfected with the CRISPR screening library, then transduced with the CAR lentivirus. Cells were expanded, then nucleofected to introduce the Cas9 protein. CRISPR-bearing cells were selected with puromycin, and CAR-positive T cells were sorted. Following extraction, genomic DNA was digested with restriction enzymes (RE), and the sgRNA-containing fragments were pulled down by biotin probe-streptavidin capture. Next, sgRNA were selectively amplified and indexed via two consecutive PCRs, and the resulting libraries were sequenced. By incorporating an intermediate step to reduce gDNA carryover, we were able to selectively retrieve and amplify our sgRNA of interest, which allows us to study the role of the kinases in CAR T cells.

Protocol

This protocol was performed in accordance with Universidad de Navarra guidelines. All subjects provided written informed consent. The reagents and the equipment used in this study are listed in the Table of Materials.

1. T cell isolation and activation

1. PBMC isolation
   1. Collect blood samples from donors on ethylenediaminetetraacetic acid (EDTA) tubes, transfer blood samples to 50 mL conical tubes, and dilute blood with PBS (dilution 1:1).
   2. Add 15 mL of density gradient medium to the bottom of a 50 mL conical tube (density for PBMCs: 1.07). Then, add carefully 25 mL of diluted blood over the density gradient medium layer (slowest speed of the pipette controller). Centrifuge for 30 min at 800 × g without brake.
   3. Collect PBMCs ring/layer with a Pasteur pipette into a 50 mL conical tube and add PBS to a total volume of 50 mL to wash. Centrifuge at 650 × g for 8 min. Discard the supernatant.
      NOTE: Check the turbidity of the supernatant to be sure that there are no cells there. If in doubt, divide the suspension into twice the original number of 50 mL conical tubes, add PBS until 50 mL, and centrifuge again.
   4. Resuspend in 50 mL of PBS. Count with acridine orange-propidium iodide (AOPI) in a cell counter. Centrifuge at 650 × g for 8 min.
2. CD4 and CD8 magnetic selection
   1. Prepare fluorescence-activated cell sorting (FACS) buffer: PBS, EDTA 2.5 µM, BSA 0.5% and filter through 0.2 µm.
   2. Resuspend PBMCs in 80 µL of FACS buffer sterile per 10 × 10⁶ cells and add 10 µL of microbeads CD4 and 10 µL of microbeads CD8 per 10 × 10⁶ cells. Incubate 20 min at 4 °C.
   3. Add 10 mL of FACS buffer and centrifuge for 8 min at 650 × g. Discard the supernatant completely.
   4. Resuspend in 500 µL per 1 × 10⁸ cells of FACS Buffer in a 15 mL conical tube.
      NOTE: A run can accommodate 200 x 10⁶ cells.
   5. Use the Possel program for AutoMacs (high-speed magnetic cell sorter) isolation.
   6. Add up to 10 mL of PBS, count with AOPI in a fluorescent cell counter, and centrifuge at 650 × g for 8 min.
   7. Resuspend cells 1 x 10⁶/mL in T cell growth medium containing 3% human serum (HS), 1% penicillin/streptomycin (P/S), 625 IU/mL interleukin (IL)-7, 85 IU/mL IL-15. Add 10 µL of T cell stimulation reagent per 1 mL and mix well. Plate 2 mL per well in a 24 well plate (p24w) and incubate for 24 h at 37 °C.

2. CAR T cell production and CRISPR screening

1. Library spinfection and CAR transduction
   1. Add the corresponding amount of lentivirus CRISPR library (e.g., Brunello Kinome 1 library. See Supplementary Table 1), determined in a titration assay, and 8 µg/mL polybrene.
   2. Spinfect by centrifugation at 700 x g for 1 h 30 min at 32 °C.
   3. Replace media with fresh T cell growth medium containing 3% HS, 1% P/S, 625 IU/mL IL-7, 85 IU/mL IL-15 (final [cell]= 10⁶ cells/mL).
   4. After 6 h, add the corresponding amount of CAR lentiviral vector to a multiplicity of infection (MOI) of 3 and incubate for 4 days.
2. Cas 9 electroporation
   1. Collect CAR T cells, wash once with PBS, and count cells with AOPI in a fluorescent cell counter.
   2. Prepare 4 µM Cas9 in electroporation buffer.
   3. Resuspend CAR T cells in 50 µL per 5 × 10⁶ cells. Mix and carefully add 50 µL of cell suspension in each well of a multiwell cuvette strip without forming bubbles. Then, using a pipette tip, remove any possible bubbles and run the tip along the edges of the well.
      NOTE: If any well in a cuvette is left without a sample, add the same volume of electroporation Buffer. This is a critical step: if the machine detects a bubble or volume difference, it will give an error.
   4. Use the program EXPAND T CELL 3 to nucleofect CAR T cells.
   5. Leave the cells in the cuvettes in an incubator at 37 °C for 40 min.
   6. Resuspend the nucleofected cells in T cell growth medium containing 3% HS, 1% P/S, 625 IU/mL IL-7, 85 IU/mL IL-15 (final [cell]= 10⁶ cells/mL) and incubate at 37 °C for 3 days.
3. Puromycin selection
   1. Count with AOPI in a fluorescent cell counter.
   2. Adjust to 1 x 10⁶ cells/mL with T cell growth medium containing IL7 and IL15.
   3. Add puromycin (2.5 µg/mL) to the media.
   4. Count cells every 2 days for 6 days and adjust to 1 x 10⁶ cells/mL with cell growth medium containing 3% HS, 1% P/S, 625 IU/mL IL-7, 85 IU/mL IL-15, and 2.5 µg/mL puromycin.
   5. Check % CAR.
4. Sort CAR T cell populations and store samples (dry cell pellet) at -80 °C.

3. Isolation of gDNA

NOTE: Avoid any cell or gDNA losses throughout the process, as this may compromise the representativity of the sgRNAs. Aim for a sgRNA coverage of at least 500x.

1. Extract gDNA from samples using the cells and tissue DNA extraction kit. Elute twice to maximize gDNA recovery.
   NOTE: Do not use more than 4 x 10⁶ cells/column in order to recover the gDNA more efficiently.
2. Measure gDNA concentration with the dsDNA quantification kit.
   NOTE: If the concentration is too low for subsequent steps in the protocol, use solid-phase reversible immobilization (SPRI) bead-based clean-up reagent to concentrate the DNA.

4. Enrichment of sgRNA cassette (and elimination of gDNA carryover)

1. Restriction enzyme digestion
   1. Digest a maximum of 5 µg gDNA with 20 U of the enzyme/s (e.g., for lentiGuide-puro backbone for Brunello Kinome 1 library use the combination NdeI and PspXI) in a total volume of 50 µL. Scale up the number of reactions to achieve the desired sgRNA representativity.
      ​NOTE: Select the restriction enzymes to cut in the flanking regions of the cassette, but bear in mind that some of the sgRNAs might contain a restriction site. These sgRNAs will be lost in this step, and they will not show up in the sequencing. Absence of this guide serves as a control of digestion. Supplementary Table 1 shows a list of Kinome 1 sgRNAs that are cleaved by PspXI and NdeI.
   2. Incubate overnight at 37 °C in a thermocycler.
   3. Perform a 2x SPRI clean-up.
      1. Prepare 10 mL of Elution Buffer (EB) (10 mM Tris-HCl, pH 8.0) and 50 mL of fresh 70% ethanol.
      2. Vortex the beads and add 100 µL to the product of the digestion.
      3. Mix thoroughly by pipetting and incubate for 5 min. Magnetize for another 5 min.
      4. While on the magnet, add 200 µL of 70% ethanol without disturbing the pellet. Wait 30 s, then discard the supernatant.
      5. Repeat step 4.1.3.3 for a total of 2 washes. Allow the beads to dry for 2 min.
      6. Remove from the magnet, add 40 µL of elution buffer, and resuspend the beads by pipetting. Incubate for 2 min.
      7. Magnetize for 2 min and transfer the supernatant to a new tube.
         NOTE: When setting up this protocol for the first time, verify the digestion through (capillary) electrophoresis. Ensure that a smear is observed.
2. Pulldown of the sgRNA cassette.
   1. Preparation of streptavidin magnetic beads.
      1. Prepare Wash/Binding Buffer 2x: 10 mM Tris-HCl pH 7.5, 2 M NaCl, 1 mM EDTA. Dilute half of the volume with H₂O to 1x.
      2. Vortex the beads for 1 min.
      3. Transfer the desired volume of beads to a tube (typically 1 mg will suffice), add an equal volume of Wash/Binding Buffer 1x, or at least 1 mL, and resuspend.
      4. Place the tube on a magnet for 1 min and discard the supernatant.
      5. Remove the tube from the magnet and resuspend the washed beads in the same volume of 1x Wash/Binding Buffer as the initial volume of beads taken from the vial.
      6. Repeat for a total of 3 washes.
      7. Resuspend beads in 2x Wash/Binding Buffer.
   2. Add 5 µL of 10 µM pulldown primers listed in Table 1 to the purified digestion from step 4.1.3.6.
   3. Incubate in a dry bath at 96 °C for 5 min, then immediately bring to ice for 5 min.
   4. Add 1 mg of pre-washed streptavidin magnetic beads.
      NOTE: Biotin-streptavidin binding works best at 1 M NaCl. Make sure to mix equal volumes of beads and biotinylated DNA.
   5. Incubate for 20 min while resuspending the mix every 4 min.
   6. Magnetize for 5 min.
   7. Wash the beads 3 times with 1× Wash/Binding Buffer.
   8. Resuspend the beads in 50 µL of EB 8.0.
      NOTE: The sgRNA cassettes used here are bound to the streptavidin beads. Do not discard them.

5. sgRNA library preparation

1. PCR1
   1. Prepare PCR1 mix with 20 µL of Herculase 5x buffer, 1 µL of dNTP (100 mM), 1 µL of DNA polymerase, 2.5 µL of 10 µM of PCR1 forward and reverse primers listed in Table 1, and 23 µL of PCR grade water. Pipette mix and add PCR1 mix to the beads from step 4.2.8.
      NOTE: Table 1 lists PCR1 primers designed for the Brunello Kinome Library 1.
   2. Place the tubes in a thermocycler and run the program for PCR1 described in Table 2.
      NOTE: Make sure the thermocycler can accommodate 100 µL reactions. Otherwise, aliquot in 50 µL reactions. Scale as needed.
   3. Pool all PCR1 from the same samples in a single tube.
   4. SPRI clean-up PCR1 using 150 µL (1.5x) of beads (follow steps 4.1.3.3-4.1.3.7). Resuspend each PCR reaction in 100 µL of EB 8.0.
      NOTE: If setting up this protocol for the first time, verify the size and quality of the PCR product by running 2 µL of a 30-cycle PCR product in a capillary electrophoresis system (e.g., Agilent's TapeStation Systems).
2. PCR2
   1. Prepare PCR2 mix with 10 µL of the product of PCR1, 10 µL of 5X buffer, 0.5 µL of dNTP (100 mM), 1 µL of DNA polymerase, 1.25 µL of 10 µM PCR2 primers forward and reverse listed in Table 1 (use a different rP7-i7-BC and P5-i5-BC combination for each sample), and 26 µL of PCR grade water for a total volume of 50 µL.
      NOTE: To maintain library representation, perform one PCR2 reaction per 10⁴ constructs in the library.
   2. Place the tubes in a thermocycler and run the program for PCR2 described in Table 2.
   3. SPRI clean-up PCR2 using 75 µL (1.5x) of beads (follow steps 4.1.3.3-4.1.3.7). Resuspend each PCR reaction in 50 µL of EB 8.0.
3. Quantify PCR2 (i.e., the sgRNA library) using the dsDNA quantification kit and verify the amplicon size by capillary electrophoresis (e.g., TapeStation Systems). Pool the samples.
4. Sequence on an Illumina instrument.
   NOTE: For the 3052 sgRNA Brunello Kinome 1 library here, a sequencing depth of 100-200 million reads per sample is typically enough.

6. Analysis

1. Demultiplex samples using bcl2fastq.
2. Check the quality of the fastq files generated with FastQC.
3. Extract the sequence of the sgRNA using as a pattern the cassette sequences flanking the sgRNA, as well as the length of the guide (20 nt).
4. Align the sequence extracted to the library reference using Bowtie2 to check how many of the hypothetical guides are authentic sgRNA from the library. Use samtools to convert the SAM file with information about the alignment into BAM format and generate summary files.
5. Count and normalize the number of reads for each sgRNA in the edited and control samples with the count function from MAGecK software.
6. Compare the counts with the test function of MAGecK to obtain the RRA scores and the list of candidate genes.

Results

As described in the Protocol Section, CAR T cells were generated following our established protocol²⁶. For CRISPR screening, we adapted the workflow developed by Wang et al.¹⁷ (Figure 1A). Briefly, T cells from two independent donors were isolated and activated for 24 hours before consecutive transduction with both the CRISPR sgRNA library and the CAR lentiviral vector, reaching 67 and 72% CAR-positive cells at day 5. A fraction of cells was nucleofected with Cas9 protein to induce knockouts, while the remaining cells served as a baseline for sgRNA representation. Transduced cells were selected with antibiotics, and distinct populations were subsequently sorted according to the phenotypes under study. CAR T cell proliferation was monitored throughout the screen using AOPI-based cell counts and expressed as fold change (Figure 1B). Following puromycin supplementation at days 8, 10, and 12, we confirmed successful enrichment of CAR T cells transduced with the CRISPR library in both donors, as evidenced by continued survival and proliferation. Untransduced T cells (UTD) and CAR T cells lacking CRISPR library transduction were used as controls of antibiotic selection.

After genomic DNA extraction, sgRNA libraries were prepared via PCR amplification. To overcome issues arising from excessive gDNA carryover, we implemented a protocol incorporating restriction enzyme digestion and biotin-streptavidin pulldown of the sgRNA cassette (Figure 2A,B). gDNA was digested with restriction enzymes NdeI and PspXI targeting sites flanking the sgRNA cassette (Figure 2C). Biotinylated oligos, designed to bind upstream of PCR1 primer annealing sites, were used to capture sgRNA cassette fragments, then pulled down with streptavidin beads to eliminate carryover gDNA. Fragments containing sgRNA were then selectively amplified (Figure 2D) and indexed through 2 consecutive PCRs.

Libraries from each sample were sequenced using Next Generation Sequencing (NGS). After demultiplexing, quality control, and sgRNA identification, data were analyzed with the MAGeCK pipeline. First, the count command quantified and normalized sgRNA read counts in control (basal) and Cas9-edited samples (Figure 3A,B). Pearson and Spearman correlations indicate that the representativity of sgRNA was maintained during the experiment (donor gDNA) compared with the plasmidic library (Figure 3C). Next, the test command compared both conditions, assigning robust rank aggregation (RRA) scores to each gene (Figure 3D). Significantly enriched or depleted genes were identified using thresholds of p < 0.05 and log fold change (LFC) > 0.5 (Figure 3E). Finally, pathway enrichment analysis of candidate genes was performed using Metascape, revealing top Gene Ontology terms associated with the identified regulators (Figure 3F).

Figure 1: CRISPR screening in CAR T cells. (A) Scheme of the CRISPR screening in CAR T cells. (B) Selection of transduced cells with puromycin (P). Plasmidic libraries include a puromycin resistance gene for positive selection. Antibiotic was added at day 8 to the media in non-transduced samples of untransduced (UTD) and CAR T cells and transduced CAR T cells. Samples without puromycin were used as a control of proliferation. Data from two independent donors is shown. Please click here to view a larger version of this figure.

Figure 2: Library preparation for sgRNA enrichment analysis. (A) Scheme of the protocol for library preparation. (B) Scheme of the relative localization of the cut sizes of the restriction enzyme (scissors) and the annealing regions of PCR1 primers (arrow) and biotinylated primers (arrow with green dot). (C) Electrophoresis of the gDNA pre- and post-RE digestion. (D) Selective enrichment of sgRNA cassettes after PCR1. Data from two independent donors from basal and Cas9 samples of the two phenotypes is shown. Please click here to view a larger version of this figure.

Figure 3: Guide enrichment analysis with MAGeCK. (A) Summary of the output of the MAGeCK count function. (B) Distribution of read counts. (C) Pearson and Spearman coefficients for correlation between donor gDNA and plasmid library counts. (D) Potential enriched/depleted genes ranked by a modified robust ranking aggregation (RRA score). (E) Volcano plot showing significant genes (p-value < 0.05, LFC > 0.5) in red. (F) Top Gene Ontology (GO) terms after Metascape pathway analysis of the enriched significant genes. Please click here to view a larger version of this figure.

Table 1: Sequence of primers. Please click here to download this Table.

Table 2: PCR conditions. Please click here to download this Table.

Supplementary Table 1: List of sgRNA in Kinome 1 library and sgRNAs lost during digestion with NdeI and PspXI Please click here to download this File.

Discussion

CRISPR screening relies on the principle of single-guide perturbation, meaning that only one sgRNA should be integrated per cell. To achieve this, sgRNAs are typically delivered via lentiviral vectors, and the multiplicity of infection (MOI) must be carefully controlled. An MOI of ca. 0.3 is generally recommended, resulting in 25% to 30% of cells being transduced with a single viral particle⁶. Accurate titration is therefore essential to determine the volume of viral particles required. Another critical step is the loss of sgRNA coverage during the protocol. Cell loss during harvesting, inefficiencies during gDNA extraction (namely column overloading and incomplete elution of DNA), or excessive template during PCR amplification can all contribute to this loss, hence skewing our sgRNA analysis. Regarding the biotinylated oligos, it is best if they are HPLC purified to avoid that free biotin used in the synthesis saturates the streptavidin beads. Moreover, a 3' amino-modifier is strongly suggested to prevent amplification from the biotinylated primers in subsequent PCRs. Also, when designing these oligos, avoid any overlap with PCR1 primers to prevent competition during amplification. In addition, bioinformatic analysis requires careful attention: sgRNA extraction patterns depend on the plasmid backbone (e.g., lentiGuide-puro or lentiCRISPR v2), and alignment stringency can range from exact matching to some tolerance of mismatches. Furthermore, since this protocol includes a digestion step, some sgRNAs containing restriction sites may be lost, requiring adjustment of the reference library.

In this study, we propose an optimized CRISPR screening library preparation protocol that enriches sgRNA-containing fragments through digestion and, more importantly, a pulldown step. Restriction enzyme digestion has been previously suggested as a strategy to reduce gDNA input and enrich for sgRNA cassettes²⁷. Similarly, biotinylated probes have been widely applied in molecular biology to capture specific nucleic acids or protein-nucleic acid complexes^28,29,30,31.

Our protocol builds on these principles and adapts them for CRISPR screening in primary CAR T cells, although several limitations should be acknowledged. First, primary CAR T cells impose constraints on screening scale, as their expansion capacity is limited. To maintain adequate coverage and sgRNA representation, large numbers of starting cells are required. When this is not feasible, smaller custom libraries may be used¹², though this reduces the breadth of genetic interrogation. Biological variability is another challenge; differences between donors are expected, and the overlap of significant hits is often modest¹⁷. Second, the enrichment strategy relies on the presence and position of restriction enzyme cutting sites. In rare cases, these sites may occur within sgRNA sequences, which is more problematic for small libraries with limited redundancy. However, most libraries, particularly genome-wide designs, include multiple sgRNAs per target, minimizing the overall impact of guide loss. Finally, alternative fragmentation-based enrichment strategies, such as sonication coupled with probe capture²⁴, may provide some flexibility in cases where restriction enzyme digestion is not suitable.

Overall, the sgRNA cassette represents a tiny fraction of the total genomic DNA in CRISPR-edited cells, making efficient enrichment critical for successful library preparation. Compared with standard protocols, our optimized workflow reduces gDNA carryover and favours PCR amplification of the sgRNA of interest. While the combination of gDNA fragmentation and biotinylated probe pulldown with streptavidin beads is not conceptually new, its integration into CRISPR screening avoids common PCR pitfalls and increases CRISPR screen reproducibility. Moreover, incorporating standardized QC metrics at each key step (such as post-sorting T-cell purity, the proportion of puromycin-resistant cells, gDNA integrity, and CAR transduction efficiency) would further enhance the robustness and reproducibility of the workflow. In conclusion, although developed in the context of primary CAR T cells and with only two donors (that may represent a limitation, although several samples from each donor were used), this optimized protocol may be broadly applicable to other CRISPR screenings and NGS workflows with large gDNA carryovers.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
4200 TapeStation	Agilent	G2991A
Acridine Orange	Invitrogen	A1301
AMPure XP	Beckman Coulter	A63881
autoMACS NEO Separator	Miltenyi	130-120-327	High-speed magnetic cell sorter
Biotinylated oligos	IDT
BSA	Sigma	A9647
Cellometer K2 Fluorescent Cell Counter	Revvity	CMT-K2-MX-150	Fluorescent cell counter
CHT4 Counting Chambers	Revvity	CHT4-SD100-002
CytoSinct CD4 Nanobeads	GenScript	L00863-1
CytoSinct CD8 Nanobeads	GenScript	L00864-1
dNTP	Agilent	200418-51
Dynabeads M-280 Streptavidin	Invitrogen	11206D
DynaMag-2 magnet	Invitrogen	12321D
EDTA 0.5 M pH 8.0	Invitrogen	15575-038
Ethanol	Merck	1,00,98,31,000
ExPERT Atx Electroporator	MaxCyte	ExPERT ATx
Ficoll-Paque	Cytiva	17544203
Filtropur S 0.2 µm	Sarstedt	83,18,26,001
GenCRISPR Ultra NLS-Cas9-Research	GenScript	Z03621-1
Herculase Buffer 5X	Agilent	600675-52	5X buffer
Herculase II Fusion Enzyme	Agilent	600679-51	DNA polymerase
IL-15	Miltenyi	130-095-765
IL-7	Miltenyi	130-095-362
Mastercycler X50a	Eppendorf	6313HG006059
MaxCyte Electroporation Buffer	Cytiva	EPB1
NdeI	New England Biolabs	R0111L
Nucleospin Tissue	Macherey-Nagel	74,09,52,250
PBS	Gibco	14190-094
Polybrene	Merck	TR-1003-6
Primers	ThermoFisher
Propidium Iodine	Sigma	P4864
PspXI	New England Biolabs	R0656S
Puromycin	Gibco	A11138-03
Qubit 3.0 Fluorometer	Invitrogen	Q33216
Qubit HS DNA assay	Invitrogen	Q32851
R-50x3 Sterile Well Processing Assembly	MaxCyte	LM243882
T Cell TransAct human	Miltenyi	130-111-160	T cell stimulation reagent
Tape Station HSD1000	Agilent	5067-5585
TEXMACS	Miltenyi	130-097-196	T cell growth medium
Tris Buffer, 1.0 M, pH 8.0, Molecular Biology Grade	EMD Milipore	648314
Water, for molecular biology, DNAse, RNAse and Protease free	ThermoScientific	327390010