A reporter system for enriching CRISPR/Cas9 knockout cells in technically challenging settings like patient models

CRISPR/Cas9 represents a valuable tool to determine protein function, but technical hurdles limit its use in challenging settings such as cells unable to grow in vitro like primary leukemia cells and xenografts derived thereof (PDX). To enrich CRISPR/Cas9-edited cells, we improved a dual-reporter system and cloned the genomic target sequences of the gene of interest (GOI) upstream of an out-of-frame fluorochrome which was expressed only upon successful gene editing. To reduce rounds of in vivo passaging required for PDX leukemia growth, targets of 17 GOI were cloned in a row, flanked by an improved linker, and PDX cells were lentivirally transduced for stable expression. The reporter enriched scarce, successfully gene-edited PDX cells as high as 80%. Using the reporter, we show that KO of the SRC-family kinase LYN increased the response of PDX cells of B precursor cell ALL towards Vincristine, even upon heterozygous KO, indicating haploinsufficiency. In summary, our reporter system enables enriching KO cells in technically challenging settings and extends the use of gene editing to highly patient-related model systems.

www.nature.com/scientificreports/ chose a dual fluorochrome based system that enables selection of transfection positive and importantly, nuclease active cells (Fig. 1A). The reporter constitutively expresses iRFP-720 (iRFP) as a fluorochrome marker to enrich transduced cells using flow cytometry. The second fluorophore of the reporter, destabilized green fluorescent protein (GFP), is cloned out of frame (+ 1), so that it is only expressed if an appropriate frameshift (− 1 or + 2) is introduced upstream, by error-prone non-homologous end joining (NHEJ) following Cas9-mediated doublestrand breaks ( Fig. 1A and Figure S1). We decided for an "on" reporter where onset of expression of the fluorochrome indicates successful editing, because "off " reporters are hampered by the time required until expression of the reporter marker is completely vanished. While we decided for a fluorochrome marker for enrichment, any other selection strategy, like a resistance gene, can also be used. Both fluorochromes and the sequence between them is translated into protein, adding an N-terminal nonsense peptide to GFP as the T2A site was cloned 3′ of iRFP and 5′ to the Cas9:sgRNA target site to prevent putative indels in the T2A site. To detect successful editing of genes of interest (GOI), we cloned distinct Cas9:sgRNA target sites between both fluorochromes ( Fig. 1A and Figure S1). Each Cas9:sgRNA target site is a copy of 20 bp of the genomic sequence of the GOI that is recognized by the specific single strand guide RNA (sgRNA), including the Streptococcus pyogenes Cas9 (SpCas9) 5′-NGG-3′ protospacer adjacent motif (PAM) sequence.
With the goal to use the reporter in PDX leukemia cells, we aimed to increase efficiency and reduce rounds of in vivo cell amplification. We advanced published reporters 6 and cloned 17 Cas9:sgRNA target sites in a row; multiplexing of different Cas9:sgRNA target sites allowed generating a single transgenic PDX line with one reporter construct enabling the studies of numerous different target genes. While attractive, the accompanying challenge lies in arranging all 17 Cas9:sgRNA target sites such as to prevent premature stop codons in the entire reporter before gene editing. A stop codon existing in any of the Cas9:sgRNA target sites might prevent translation of GFP upon successful gene editing at a neighboring Cas9:sgRNA target site. As a consequence and prerequisite, each individual Cas9:sgRNA target site required absence of stop codons in at least one reading frame, while adding 1 or 2 bases between Cas9:sgRNA target sites allowed having each Cas9:sgRNA target site in the desired frame lacking stop codons ( Figure S1A).
The reporter contains a non-human targeting sequence (Gaussia Luciferase, GLuc) at position 2 ( Fig. 1A and Supplemental Table S1), which is exclusively present in the reporter and serves as a control; the remaining 16 targeting sequences can be used to select edited cells of distinct GOIs. At position 16, the reporter contains the LYN genomic targeting sequence ( Fig. 1A and Supplemental Table S1).
To avoid undesired interference, we added a glycine linker between the most 3' Cas9:sgRNA targeting site and GFP ( Figure S2A) and introduced distancing between GFP and the nonsense peptide translated from the Cas9:sgRNA target sites. As additional advantage, the linker prevented destruction of the coding region of GFP upon large deletions by NHEJ which might increase the number of false-negative reporter cells 15,16 .
We first used a cell line system to test different aspects of the reporter and test its suitability. We chose a B-ALL cell line, NALM-6, which stably expressed human codon-optimized S. pyogenes Cas9 ( Figure S2B). To test different lengths of linker sequences, Cas9-expressing NALM-6 cells were transduced with reporter constructs containing either no linker or an 15 or 48 bp linker ( Figure S2A), and were sorted for iRFP. In a second round, cells were transduced with either an empty sgRNA (EV-sgRNA, Figure S2B) vector expressing mTagBFP as control, an ADAM17-targeting or a LYN-targeting sgRNA vector (expression constructs are detailed in Fig. 1A, Figure S2 and Supplemental Table S1). Compared to cells that contained the reporter without linker, the short 15 bp linker already increased marker expression within the fraction of sgRNA-expressing cells. The 48 bp linker The reporter consists of two fluorochromes, iRFP720 and GFP, both expressed from a single promoter (EF1α) and coupled by the self-cleaving peptide T2A. While iRFP720 is constitutively expressed, the GFP cassette is cloned out-of-frame by + 1 base. In the 5′ untranslated region of GFP, 17 different Cas9:sgRNA target sites were cloned in, each consisting of a copy of the genomic target sequence of the gene of interest (GOI) including its PAM sequence; the Cas9:sgRNA target sites were followed by a 48 bp glycine-rich linker (dashed line). Upon co-expression of Cas9 and sgRNA, Cas9 induced DNA double strand breaks in the specific Cas9:sgRNA target site of the reporter; non-homologous end joining (NHEJ)-mediated frame shifts (Indels), which altered the reading frame by − 1 or + 2, enabled expression of GFP. (B) Gating strategy; NALM-6 cells stably expressing Cas9 and the reporter were lentivirally transduced mock (without sgRNA construct), a GLuc-targeting sgRNA or a LYN-targeting sgRNA construct that co-expressed mTagBFP ( Figure S2  www.nature.com/scientificreports/ was even more potent and enriched ADAM17-sgRNA or LYN-sgRNA and GFP double-positive cells to more than 55% and 24%, respectively ( Figure S2C); false-positive rate was negligible, with only 0.77% GFP-positive cells in the EV-sgRNA control cells ( Figure S2C). Thus, the final reporter used for all subsequent experiments contains the 48 bp linker.
To test the feasibility of the reporter in enriching genome-edited cells, Cas9 and reporter co-expressing NALM-6 cells were transduced without or with either GLuc control or LYN-targeting sgRNA lentiviral vectors (Fig. 1B); to mimic the condition of viral transduction of primary samples, and to aim for a single integration of the lentivirus in the genome 17 , cells were transduced with a low transduction efficiency. Three days after infection, cells were sorted by flow cytometry for either mTagBFP-iRFP double positive or, indicative for Cas9-induced frame shift mutations in the reporter gene, mTagBFP-iRFP-GFP triple-positive cells (Fig. 1B). While around 20% or 40% of cells expressed the sgRNA in control and LYN sgRNA transduced cells, respectively, the reporter was edited in less than 3% of the LYN KO cell population, as indicated by GFP expression (Fig. 1B). Minor reporter positivity in cells with both Cas9 and the sgRNA in part reflects incomplete gene editing, but also the fact that the reporter is restricted to recognize Indels with a frame shift of − 1 or + 2, while other Indels remain undetected. Capillary immunoassay revealed a dramatic decrease of LYN protein expression in GFP-positive cells as compared to GFP-negative LYN-sgRNA-expressing cells or cells expressing the control GLuc-sgRNA (Fig. 1C). Importantly, targeting genes located at distinct positions within the reporter demonstrated similar enrichment, indicating that editing efficiency is independent of the position of the targeting sequence within the reporter ( Figure S3).
As a next step, we performed droplet digital PCR (ddPCR) to evaluate the percentage of genome-edited alleles at the LYN locus in each sub-population (Fig. 1D, Figure S4, Supplemental Table S2) 18 . ddPCR detected a significant enrichment of edited alleles in the reporter-positive subpopulation to around 80% as compared to the reporter-negative population, which contained below 10% edited alleles (Fig. 1D), indicating that cells with gene editing in the reporter revealed reliable gene editing in the genome. These data demonstrate that the reporter system reliably identifies cells with successfully edited loci, enabling their enrichment.
To gain insight into the editing efficiency on a single cell level, Cas9 and reporter co-expressing NALM-6 cells were electroporated with a LYN-gRNA (see Methods for details) and single cell cultures were generated from GFP-positive and GFP-negative cells. Electroporation was used to prevent ongoing Cas9 activity, and putatively associated off target effects 19,20 , typical for lentivirally transduced cells with genomic transgene integration. Protein expression analysis demonstrated that 48% (10/21) of GFP-negative clones showed decreased LYN protein expression, while reduced expression of LYN was detected in 95% (21/22) of GFP-positive clones ( Fig. 1E and Figure S5A); in 5/22 LYN protein levels were markedly decreased but still readily detectable, which may point to a heterozygous knockout in these clones. Most reporter-positive clones (16/22) demonstrated complete loss of the protein, indicating homozygous gene deletion ( Fig. 1E and Figure S5A). Thus, the reporter increased efficiency and reduced the resources required for generating single cell clones, even in straightforward cell lines systems.
To evaluate a functional role for LYN in B-ALL, we expanded 3 of the single cell clones ( Figure S5A, bold coloured labels) with putative homozygous (hom) or heterozygous (het) knockout or wildtype (wt) LYN genotype and confirmed genotypes by ddPCR ( Figure S5B) and protein expression analysis (Fig. 1F, inner panel). In competitive growth assays, no difference in growth behavior between the 3 genotypes was detected after 2 weeks ( Figure S6A), indicating that LYN is dispensable for proliferation and survival of NALM-6 cells in vitro. In a next step, we tested response of the single cell clones to conventional chemotherapeutic drugs and detected enhanced response of LYN-edited cells to VCR (Fig. 1F), as well as methotrexate and etoposide treatment ( Figure S6BC) as compared to GLuc controls (ctrl). Interestingly, IC50 of VCR was equally reduced in LYN homozygous and heterozygous knockout cells, as compared to wildtype or GLuc control cells (Fig. 1F). Similarly, a 50% reduction of LYN protein expression in the heterozygous cells equally sensitized cells to the additional chemotherapeutic drugs methotrexate and etoposide as compared to LYN-knockout cells ( Figure S6BC). Importantly, the increased drug-sensitivity in the single cell clones was phenocopied when the bulk population of reporter-positive cells was analyzed as compared to GLuc controls, reflecting the high enrichment rate of LYN-knockout cells from the bulk by the reporter ( Figure S7). Thus, LYN is an important determinant of response to chemotherapy in NALM-6 cells and even moderate reduction of LYN sensitizes NALM-6 cells to chemotherapy.
Having established the surrogate reporter, we next tested our system in two PDX ALL models, ALL-199 and ALL-256 (patient characteristics are detailed in Supplemental Table S3 as published previously [21][22][23] ). Compared to leukemia cell lines, PDX ALL cells are substantially more challenging in handling as they are hard to transduce and reluctant to grow in vitro 21,22 . We sequentially transduced PDX cells first with a Cas9 expression plasmid and second with the reporter, each followed by one round of enrichment and amplification of transduced cells in mice (constructs are detailed in Figs. 1A and S2B). We have previously analyzed clonal stability of PDX models after lentiviral transduction and amplification in mice, and did not find evidence of major clonal selection 24 , suggesting that the sequential enrichment steps may not drastically affect the overall mutational pattern of PDX samples. In a third round, PDX cells were transduced with the LYN sgRNA or control GLuc sgRNA vectors. 5 days after sgRNA transduction, PDX cells were enriched for the scarce population of reporter positive cells, which were re-transplanted into next recipient mice (Figs. 2A, S2B, S8). GFP-positive LYN and GLuc sgRNA expressing PDX cells displayed a similar expansion rate in mice (Supplemental Table S4), indicating that LYN does not significantly affect growth of PDX cells in vivo in accordance with the in vitro observation in NALM-6 cells.
PDX cells were re-isolated after passaging and frequency of genome-edited alleles and LYN protein expression was analyzed in GFP-positive cells of LYN KO compared to GLuc control PDX cells. These experiments revealed that the LYN locus was edited in around 80% of cells (Fig. 2B), which is associated with a substantial decrease of LYN protein expression (Fig. 2C), demonstrating efficient enrichment of edited PDX cells by the use of the surrogate reporter. In a next step, therapy response of PDX cells was evaluated in short term in vitro cultures. www.nature.com/scientificreports/ www.nature.com/scientificreports/ Compared to control GLuc KO PDX cells, LYN KO PDX cells were significantly more sensitive to VCR treatment in both ALL PDX models tested (Fig. 2D), reproducing the phenotype of NALM-6 cells in two PDX ALL models. Our study provided a clinically relevant genetic model to study LYN, a SRC-family kinase that is often activated in B-ALL by rare genetic and non-genetic mechanisms and has been associated with treatment response [8][9][10][11][12][13][14][15]25 . Our data indicate that LYN regulates therapy response of preB-ALL cell lines and PDX cells. Further studies are needed to evaluate whether available drugs that target LYN 26,27 can improve response to conventional therapy in relapsed B-ALL.
Taken together, our reporter system strongly enriches cells with successful Cas9-induced gene-editing from the large pool of non-edited cells. By using the surrogate reporter, we successfully selected KO clones and extended the use of CRISPR/Cas9 gene editing to PDX cells. Despite the fact that the reporter system may overestimate editing efficiency, e.g. due to differences in chromatin state at the genomic loci versus the targeting site in the reporter 28 , we demonstrate here that our reporter was highly specific to enrich for true KO cells. Application of the reporter system will enable extending the use of CRISPR-Cas9-mediated gene editing to cells which are difficult to handle, but represent highly clinically relevant and precious preclinical model systems.

Materials and methods
Plasmid cloning. The reporter plasmid is based on a pCDH lentiviral vector which expressed iRFP720 followed by T2A under control of the Elongation factor 1-alpha (EF1α) promoter. The fragment contains several Cas9:sgRNA target sequences (see supplementary Table S1, right panel) followed by a GC-rich sequence (GGGGSLVPRGSGGGGS) and was purchased from IDT Inc. (Coralville, IA, USA); in its 3′ region, the fragment contains the sequence of destabilized GFP 29 with a frame shift of + 1 base. The fragment was cloned into the pCDH vector using NotI and SalI. Details on the sequences are described in Figure S1.
Cloning of the the pCDH-based lentiviral sgRNA expression vector, with mTagBFP or mCherry expressed under the control of the EF1α promoter was previously described 23 . The sgRNA sequences (see supplementary  Table S1, left panel) were synthesized (Eurofins MWG-Operon, Ebersberg, Germany), annealed and ligated into the vector using BbsI.
For Cas9 expression, a DNA fragment containing the cDNA of spleen focus forming virus (SFFV) promoter (Addgene plasmid #27348), humanized Staphyococcus pyogenes Cas9 (hSpCas9 PX330, Addgene plasmid #42230) and a T2A-linked mTaqBFP fluorochrome (pTagBFP-C vector, PF171, Envrogen, Lawrenceville, NJ) or a truncated form of the human nerve growth factor receptor (hNGFR) was amplified by PCR and cloned into the pCDH vector (Addgene Plasmid #72263) as previously described 23 . Ethical statements. Prior to obtaining the two primary B-ALL patient samples, written informed consent was obtained from all patients or from parents/caregivers in cases in which patients were minors. The study was performed in accordance with the ethical standards of the responsible committee for human experimentation

Lentiviral transduction and enrichment of NALM-6 and PDX ALL cells. PDX ALL-265 and
ALL-199 were established and the NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mouse model performed as described [21][22][23] ; Lentiviruses were produced and cells infected as described [21][22][23] ; in PDX and NALM-6 cells, when creating the stable LYN KO cell lines, transduction efficiency was at or below 40% to aim for single integrations of lentiviruses into the host genome 17 . After lentiviral transduction, PDX cells were injected intravenously into NSG mice for amplification 24 h post infection, or maintained in vitro 4 days and enriched for transgene expression before injection. NALM-6 or re-isolated PDX cells from the 2 donor mice (Gluc and LYN) were analyzed and sorted by flow cytometry for fluorochrome expression (LSRII and FACSAria III from Becton Dickinson San Jose, CA, USA) or enriched using magnetic cell separation (MACS) targeting NGFR (Miltenyi Biotech, Bergisch Gladbach, Germany) as previously described 21,22 . Transient gRNA transfection and single cell colony. For  www.nature.com/scientificreports/ 44 µM. The mixture was heated to 95 °C for 5 min and cooled down to room temperature for at least 60 min. For electroporation, 1.2 million Cas9 and reporter co-expressing cells were re-suspended in 120 µl R buffer and 4.9 µl of 44 µM gRNA mixture were added. 100 µl were subjected to each electroporation using the Neon device (Invitrogen, Eugene, OR, USA) under optimized conditions (1350 V, 10 ms, 3 pulses). 5 days after transient transfection of the LYN gRNA, single cells were sorted each in one well of a 96-well plate by FACSAria III according to GFP expression. Cells were allowed to expand for 2-3 weeks before they were transferred to 12-well plates and used for further analysis.
Capillary protein immunoassay for protein expression analysis. Flow cytometry-enriched cell populations were incubated in lysis buffer (#9803, Cell Signaling Technology, Boston, USA) and PMSF (1:200) on ice for 30 min. Protein concentration was measured by BCA assay (#7780, New England Biolabs, Beverly, USA) and abundance of specific proteins determined by capillary immunoassay. Procedures were performed following manufacturer's instructions; in brief, each capillary was loaded with protein lysate and electrophoresis was performed; capillaries were processed to attach all proteins to the capillary wall and incubated with a single antibody; results were measured as emission curves from each capillary and "Western-Blot-like presentations" calculated thereof using the Compass software (ProteinSimple), including quantification. Final "Western-Blotlike presentations" appear clearly different from conventional Western Blots as described 23 ; due to very high sensitivity, equal loading is hard to achieve, but also not required, as protein amounts are calculated with high sensitivity and reliability. Primary antibody against LYN was purchased from R&D systems (AF3206, Minneapolis, MN, USA), FLAG from R&D systems (MAB8529, Minneapolis, MN, USA) and β-ACTIN from Novus biologicals (NB600-501SS, Littleton, CO, USA).
Statistical analysis. Two-tailed Student's t test was used to compare the difference of groups. P value less than 0.05 was considered statistically significant.