(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Efficient and error-free fluorescent gene tagging in human organoids without double-strand DNA cleavage

['Yannik Bollen', 'Molecular Cancer Research', 'Center For Molecular Medicine', 'University Medical Center Utrecht', 'Utrecht University', 'The Netherlands', 'Oncode Institute', 'Utrecht', 'Medical Cell Biophysics', 'Techmed Centre']

Date: 2022-02 

CRISPR-associated nucleases are powerful tools for precise genome editing of model systems, including human organoids. Current methods describing fluorescent gene tagging in organoids rely on the generation of DNA double-strand breaks (DSBs) to stimulate homology-directed repair (HDR) or nonhomologous end joining (NHEJ)-mediated integration of the desired knock-in. A major downside associated with DSB-mediated genome editing is the required clonal selection and expansion of candidate organoids to verify the genomic integrity of the targeted locus and to confirm the absence of off-target indels. By contrast, concurrent nicking of the genomic locus and targeting vector, known as in-trans paired nicking (ITPN), stimulates efficient HDR-mediated genome editing to generate large knock-ins without introducing DSBs. Here, we show that ITPN allows for fast, highly efficient, and indel-free fluorescent gene tagging in human normal and cancer organoids. Highlighting the ease and efficiency of ITPN, we generate triple fluorescent knock-in organoids where 3 genomic loci were simultaneously modified in a single round of targeting. In addition, we generated model systems with allele-specific readouts by differentially modifying maternal and paternal alleles in one step. ITPN using our palette of targeting vectors, publicly available from Addgene, is ideally suited for generating error-free heterozygous knock-ins in human organoids.

Funding: This work is part of the Oncode Institute, which is partly financed by the Dutch Cancer Society. HGJS received European Research Council (ERC) starting grant (IntratumoralNiche), project number 803608 ( https://erc.europa.eu/funding/starting-grants ) and NWO TOP. YB was supported by a strategic alliance between University of Twente and UMC Utrecht on Advanced Biomanufacturing (to LWMMT and HJGS). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability: The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper and its Supporting Information files. All algorithms used for the mapping [ gatk.broadinstitute.org ], mutational calling [ https://github.com/ToolsVanBox/NF-IAP ], and filtering of mutations [ https://github.com/ToolsVanBox/SMuRF , https://github.com/hartwigmedical/gridss-purple-linx ] are publicly available. Raw FCS files are available on the FlowRepository database ( flowrepository.org ) and accessible using the repository ID FR-FCM-Z4PJ.

In cell lines, large knock-ins have been generated without introducing DSBs by using the partially inactivated Cas9 D10A nickase variant [ 8 – 10 ], which generates single-strand DNA breaks (nicks) in the genomic strand that hybridizes with the guide RNA [ 11 ]. By simultaneously nicking the genomic target locus and the extremities of both homology arms within the targeting vector, a strategy known as in-trans paired nicking (ITPN) [ 8 ], efficient knock-in alleles can be generated without double-strand DNA cleavage. Unlike conventional CRISPR/Cas9-mediated genome editing, ITPN modifies target loci with high fidelity, since single genomic nicks are rarely mutagenic [ 8 , 12 ]. By avoiding double-strand DNA cleavage, ITPN enables the insertion of heterozygous reporters or pathogenic (germline) mutations with intact “untargeted” secondary alleles and with minimal risk of off-target indels. Consequently, knock-in cells can be pooled to expedite the expansion and, thus, the “generation time” of a knock-in line (2 weeks). Pooling successfully targeted organoids is particularly useful for organoid models where clonal selection is laborious. Furthermore, by avoiding clonal selection, preexisting genetic diversity in tumor-derived organoid lines is largely preserved. Here, we investigate the efficiency and fidelity of fluorescent gene tagging via ITPN in human organoids. In addition, we present a palette of easy-to-use targeting vector backbones and protocols for N- or C-terminal fluorescent gene tagging using ITPN.

Organoids, in particular of human origin, represent next-generation model systems that recapitulate in vivo tissue architecture and functionality more accurately than 2D cell lines [ 2 ]. However, the precise engineering of large knock-in reporters in organoids can be laborious when using conventional CRISPR-mediated strategies to stimulate homology-directed repair (HDR) [ 3 – 5 ] or nonhomologous end joining (NHEJ) [ 6 ] based editing. While generally effective, these strategies rely on the generation of genomic double-strand breaks (DSBs) by CRISPR-associated nucleases, which often result in both on- and off-target indel mutations as a consequence of error-prone repair by repeated cycles of NHEJ. On-target indels are often generated in the “untargeted allele” that is not carrying the knock-in and may result in missense or nonsense mutations. In addition, while HDR generally results in error-free repair, generating knock-ins via NHEJ-based ligation of a linearized DNA fragment often results in indels within the up- and downstream junctions of the knock-in allele [ 6 , 7 ]. Consequently, existing knock-in protocols inherently require sequence verification of individually picked organoid clones, which is laborious, time consuming, and eliminates genetic heterogeneity in tumor-derived organoid models.

Since the development of efficient genome editing technology, molecular and cell biological research increasingly relies on genetically modified in vitro model systems. In particular, the visualization of endogenous proteins using fluorescent knock-in reporters allows for a precise assessment of their subcellular localization and dynamics during cellular homeostasis and disease [ 1 ].

Results

To probe the efficiency of fluorescent knock-ins in human organoids (Fig 1A), we designed an N-terminal mScarlet knock-in at the human SEC61B locus. We constructed different targeting vectors in order to compare editing efficiencies of various knock-in strategies (Fig 1B). To stimulate editing via NHEJ-mediated ligation of a linearized mScarlet-coding fragment into the Cas9-generated genomic DSB [6,7], we constructed a vector carrying the mScarlet-coding sequence flanked by copies of the genomic Cas9 target site. Alternatively, we included 20 bp microhomology to stimulate genomic integration via the microhomology-mediated end joining (MMEJ) pathway [13]. In addition, we generated vectors with 1 kb homology arms following a traditional targeting vector design that is without flanking Cas9 target sites, or with flanking Cas9 target sites to support genomic integration via ITPN or in-trans paired cleavage (ITPC) [14,15].

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Fluorescent gene tagging in human organoids without double-strand DNA cleavage. (A) Schematic representation of the workflow used to capture fluorescent knock-in efficiencies in human organoids. To ensure optimal outgrowth post-electroporation, organoids are trypsinized to a cell suspension consisting of approximately 5 cell clumps. After electroporation, cells are allowed to expand for 10 days without selecting for cells that received the knock-in constructs. Prior to flow analysis, organoids may consist of partial knock-in populations. To capture the overall knock-in efficiency, organoid cultures are trypsinized to a single-cell suspension and flow analyzed. (B) Schematic representation of the SEC61B targeting strategy. mScarlet was flanked with homology arms matching up- and downstream sequences of the N-terminus of the human SEC61B locus, coding for Protein transport protein Sec61 subunit beta. Cas9 was targeted close to the start of the coding region using a gRNA as indicated (green arrow). Cas9 cleavage sites (triangles) and protospacer adjacent motifs (black bar) are indicated. Up- and downstream homology is represented in blue and red, respectively. Compositions of targeting vectors supporting different knock-in strategies are indicated. (C) Knock-in efficiencies of mScarlet at the human SEC61B locus in a patient-derived tumor organoid model using various knock-in strategies. WT or D10A nickase (Nick) SpCas9 was codelivered with targeting vectors indicated in (B). Targeting vectors were electroporated at equimolar ratios between conditions to correct for differences in vector size. Editing efficiency (% mScarlet+ cells) was determined by single-cell flow analysis 10 days post-electroporation (n = 3 independent experiments). * p < 0.05 in a Ratio paired t test. Error bars indicate SEM. The inset shows representative stills of mScarlet-SEC61B localization in patient-derived tumor organoids (scale bar = 10 μm). Underlying data are provided in S1 Data. (D) Knock-in efficiency of mScarlet at the human SEC61B locus in tumor organoids using targeting vectors with different homology arm lengths flanked by Cas9 target sites and codelivered with Cas9 D10A nickase to support ITPN (n = 3 independent experiments). Targeting vectors were electroporated at equimolar ratios between conditions to correct for differences in vector size. Error bars indicate SEM. Underlying data are provided in S1 Data. (E) As in (B), schematic showing the targeting strategy for ITPN-mediated integration of mScarlet (0.7 kb) or mScarlet-P2A-PuromycinR (1.4 kb) at the C-terminus of the human HIST1H2BC locus, coding for Histone H2B type-1C. (F) Knock-in efficiency of mScarlet (mSC; 0.7 kb) or mScarlet-P2A-PuromycinR (mSC-PR; 1.4 kb) in tumor organoids at the C-terminus of the human HIST1H2BC locus (n = 3 independent experiments). Targeting vectors were electroporated at equimolar ratios between conditions to correct for differences in vector size. The difference between mSC and mSC-PR was nonsignificant in a Ratio paired t test. Error bars indicate SEM. Underlying data are provided in S1 Data. (G) Knock-in efficiency of an mScarlet knock-in at the SEC61B locus in human colon normal and tumor organoids via ITPN using 1 kb homology arms (n = 3, n = 6 independent experiments for normal and tumor organoids, respectively). In all control conditions, the targeting vector was cotransfected with a guide targeting a different gene. The difference between normal and tumor KI organoids was nonsignificant in a two-sided unpaired t test. Error bars indicate SEM. Underlying data are provided in S1 Data. Raw FCS files are available on the FlowRepository (FR-FCM-Z4PJ). ITPC, in-trans paired cleavage; ITPN, in-trans paired nicking; MMEJ, microhomology-mediated end joining; NHEJ, nonhomologous end joining; WT, wild-type. https://doi.org/10.1371/journal.pbio.3001527.g001

Targeting vectors were coelectroporated with wild-type or D10A nickase SpCas9 expression constructs in a patient-derived tumor organoid model obtained from a colorectal cancer biobank [16]. We visually confirmed the expected localization of mScarlet within knock-in organoids for each condition prior to flow analysis of mScarlet+ cells 10 days post-electroporation (Fig 1C). Flanking homology arms with Cas9 target sites to stimulate ITPN or ITPC resulted in substantially higher editing efficiencies when compared to a traditional targeting vector design with the same homology arm length (Figs 1C and S1A). In addition, NHEJ and MMEJ conditions underperformed when compared to targeting vectors with long homology arms, in particular when combined with nickase Cas9. Notably, ITPN resulted in a similar fraction of knock-in cells when compared to a traditional knock-in strategy that uses wild-type Cas9 and targeting vectors without flanking Cas9 target sites.

To investigate the fidelity of ITPN-mediated fluorescent knock-ins, we performed sequence analyses on polyclonal knock-in lines that were generated according to above-described conditions. To determine the risk for off-target indels, we analyzed the fidelity of the secondary allele that is not carrying the knock-in as a proxy for the likeliest candidates for off-target modifications. Using TIDE analysis [17], we show that wild-type Cas9 conditions result in a high frequency of indels within the secondary allele, whereas knock-in organoids generated via ITPN displayed >99% sequence integrity of their secondary allele (S2 Fig). Next, to investigate the fidelity of ITPN-mediated knock-ins, we generated 11 clonal knock-in lines from the ITPN condition and examined the knock-in alleles via Sanger sequencing. All knock-ins contained intact 5′ and 3′ junctions and no evidence for tandem integration was found (S3A Fig). Moreover, in agreement with previous TIDE analysis on polyclonal cultures, we confirmed the absence of indels in the untargeted allele of heterozygous clones (S3B Fig). Finally, to exclude the presence of off-target editing, we performed whole genome sequencing (WGS) on 3 out of the 11 clonal ITPN-mediated knock-in lines. We investigated the somatic mutation burden of these clones in 166 regions, which were predicted in silico to be likely off-target protospacer loci. No genomic aberrations were identified in the unmodified allele, the predicted off-target protospacer regions or the 200 bases surrounding the predicted sites (S1 Table). The lack of all variants ranging from single base substitutions to structural variation breakpoints confirms the absence of mutations due to incorrectly repaired off-target nuclease activity as well as off-target integrations of the knock-in cassette.

Collectively, these data indicate that ITPN enables highly efficient and indel-free fluorescent gene tagging in human organoids and makes sequence confirmation of clonal lines unnecessary. Consequently, all knock-in organoids can immediately be pooled to expedite the expansion of the edited organoid line and to maintain genetic diversity of patient-derived tumor organoid models.

Traditional design of targeting vectors requires long homology arms to maximize the chance of homologous recombination between the genomic locus and targeting vector. However, vectors with long homology arms are challenging to assemble and are inconvenient for locus-specific genotyping by PCR. To investigate whether efficiency of fluorescent gene tagging is lost when ITPN is mediated by shorter homology arms, we generated a series of targeting vectors with decreasing homology. At the SEC61B locus, the homology demand of ITPN-mediated mScarlet integration peaked at 800 bp (Figs 1D and S1B). While vectors with shorter homology arms were accompanied with lower editing efficiencies, they were sufficient to generate knock-in lines and may be preferred in situations of challenging vector assembly and/or genotyping.

Next, to probe whether knock-in size influences editing efficiency, we designed a C-terminal knock-in at the HIST1H2BC locus and constructed targeting vectors with 500 bp homology to integrate either mScarlet (0.7 kb) or mScarlet-P2A-Puro (1.4 kb) (Fig 1E). Surprisingly, we found no substantial difference in knock-in efficiency between the 2 targeting vector variants, suggesting that a knock-in size in the range of <1.4 kb has no notable influence on editing efficiency via ITPN (Figs 1F and S1C).

Since SEC61B and HIST1H2BC are ubiquitously expressed genes, we decided to investigate if we could knock-in mScarlet-P2A-Blast into normal human colon organoids at the C-terminus of KRT20, which is exclusively expressed in differentiated cells. Following a short pulse of Blasticidin selection, we observed clonal organoids with a subpopulation of cells showing the expected cytoplasmic red fluorescence (S4 Fig). Since differentiated cells do not form organoids as efficiently as stem cells, lines that involve fluorescent knock-ins in differentiation genes such as KRT20 are best generated either using a short pulse of selection or by manually picking and pooling clonal organoids that contain (some) fluorescent cells.

Finally, we compared the efficiency of an N-terminal mScarlet knock-in at the SEC61B locus between tumor and normal colon organoids (Figs 1G and S1D). The knock-in efficiency in tumor organoids was higher (although not significant), which may be attributed to a difference in culture conditions and electroporation efficiency.

A major downside of generating targeting vectors with homology arms flanked by Cas9 target sites at their extremities is the time-intensive molecular cloning. To expedite the cloning of targeting vectors for fluorescent gene tagging at either the N- or C-terminus, we generated a series of minimalistic targeting vector backbones allowing seamless one-step integration of both homology arms using SapI-based Golden Gate assembly [4] (Fig 2A). Targeting vector backbones carrying state-of-the-art monomeric fluorescent proteins are made available from Addgene, including optional P2A-linked selection elements (Fig 2B). Using our optimized vector backbones, targeting vectors can be assembled in the same amount of cloning time as is required for the insertion of gRNA oligos into Cas9 expression vectors. Consequently, when using our vector backbones for ITPN, fluorescent reporter alleles in cell lines and organoid models can be generated in as little as 2 weeks, including molecular cloning procedures for vector assembly (Fig 2C). We summarized our recommendations for knock-in design and one-step targeting vector assembly in a protocol (S1 File). In addition, new variants of targeting vector backbones, e.g., replacing the donor with a different fluorescent protein sequence, can be generated in a short amount of time (S2 File).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. One-step targeting vector assembly and ITPN expedite fluorescent gene tagging. (A) Schematic outline of one-step TV assembly via SapI-based golden gate-mediated homology arm ligation. Homology arms can be amplified from genomic DNA or ordered as commercially synthesized DNA fragments. (B) Overview of knock-in backbone constructs available from Addgene. Knock-in backbones contain one of 4 different fluorescent proteins and optional P2A-linked resistance cassettes. Backbone constructs are suitable for knock-ins at either the C- or N-terminus, as indicated. (C) Schematic workflow outlining fluorescent gene tagging in organoids using ITPN. Following electroporation, organoids generally require approximately 10 days of outgrowth before FACS purification of fluorescent knock-in cells. Alternatively, fluorescent clonal organoids can be handpicked and pooled. Sequence verification of individually picked clonal lines is not required when editing via ITPN. ITPN, in-trans paired nicking; TV, targeting vector. https://doi.org/10.1371/journal.pbio.3001527.g002

To probe the efficiency of ITPN using our newly designed targeting vectors, we generated triple fluorescent knock-ins by simultaneous targeting of 3 separate genomic loci. Specifically, we targeted the C-terminus of the HIST1H2BC locus to knock-in mTurquoise2-P2A-puromycinR, the C-terminus of the CDH1 locus to knock-in mScarlet, and as a third locus, we included an N-terminal knock-in of mNeongreen at either the LMNA, SEC61B, or MAP4 locus (Fig 3A). DNA cocktails containing different combinations of targeting vectors and their respective Cas9 expression constructs were electroporated into fractionated tumor organoids. Organoids were allowed to form for 10 days without puromycin selection prior to quantification of the raw knock-in efficiencies by single-cell flow analysis. As expected, in all 3 conditions, the knock-in fractions were dominated by cells that carried single knock-ins in either one of the targeted genes. However, we readily detected cells carrying multiple knock-ins, including cells where all 3 genes were edited simultaneously (Fig 3B). The overall knock-in efficiencies for each targeted gene and the fraction of cells carrying multiple knock-ins are summarized in Fig 3C. To confirm the fidelity of the gene fusions, we generated polyclonal triple knock-in lines from each editing condition by manual picking and pooling clonal triple positive organoids. TIDE analysis again confirmed the absence of on-target indels in the untargeted alleles of all edited genes (S5 Fig). In addition, we confirmed the intended integration of the knock-in via Sanger sequencing (S6 Fig). Next, we recorded overnight growth of our TKI-3 knock-in line using live-cell imaging to demonstrate normal growth behavior and phenotype (Fig 3D). Each channel could be recorded without excessive bleaching, allowing a multidimensional dynamic readout of chromosomal instability during mitosis, including chromatin errors (H2B1C), spindle assembly (MAP4), and membrane defects or binucleation (CDH1).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Multiplexed fluorescent gene tagging in human organoids using ITPN. (A) Multiplexed fluorescent gene tagging in tumor human colon organoids at 3 different genomic loci using ITPN. C-terminal integrations of mScarlet at the CDH1 locus and mTurquoise2-P2A-Puromycin into the HIST1H2BC locus were combined with N-terminal integration of mNeongreen at either the LMNA, SEC61B, or MAP4 locus. In the schematics: Cas9 D10A nick positions (red triangles) and protospacer adjacent motifs (black bars) are indicated for each knock-in design, as well as the gRNA used (green arrow). Organoids were electroporated simultaneously with all 3 targeting vectors to generate one-step multiplexed triple knock-ins. (B) All 3 targeting combinations yielded triple knock-in populations with practical efficiencies, as indicated by flow analysis (numbers indicate frequencies (%) of knock-in cells within the entire targeted cell population). Imaging snapshots show the expected subcellular localization of each fusion protein (scale bar = 10 μm). Raw FCS files are available on the FlowRepository (FR-FCM-Z4PJ). (C) Overview of the multiplexed gene tagging efficiencies as determined by flow cytometry analysis. Raw FCS files are available on the FlowRepository (FR-FCM-Z4PJ). (D) Live-cell imaging of tumor human colon organoids carrying CDH1-mScarlet, HIST1H2BC-mTurquoise2, and mNeongreen-MAP4 knock-ins. The top panel shows representative stills of organoid growth over time (scale bar = 10 μm). For divisions I and II, snapshots of each channel are shown in metaphase and anaphase (scale bar = 5 μm). ITPN, in-trans paired nicking. https://doi.org/10.1371/journal.pbio.3001527.g003

Taken together, these results demonstrate that ITPN maintains high levels of fidelity across different genomic loci and allows multiplexed fluorescent gene tagging in human organoids. Using conventional editing protocols, generating organoid lines carrying multiple fluorescent knock-ins is highly laborious. By using ITPN, organoids with multiple edits can be generated within 2 weeks. Alternatively, in case an attempt to multiplex gene targeting fails, cells with a single knock-in can be pooled and retargeted. Moreover, we generated the same combinations of triple knock-ins in 2 rounds of targeting and used intermediate antibiotic selection to enrich for knock-in cells instead of manual picking (S7 Fig).

Since the sequence integrity of the untargeted allele that is not carrying the knock-in is maintained when editing via ITPN, this secondary allele can be retargeted using the same locus-specific targeting vector to obtain homozygous knock-ins. This also enables straightforward differential modification of maternal and paternal alleles by offering 2 different targeting vectors for the same locus. To investigate if ITPN allows the simultaneous generation of biallelic knock-ins carrying different fluorescent tags within each allele, we targeted the SEC61B, MAP4, and HIST1H2BC loci in tumor organoids with both mNeongreen and mScarlet targeting vectors. Flow analysis at 10 days post-electroporation confirmed the presence of a double-positive cell population for each targeted locus (Fig 4A). Genotyping of manually picked lines confirmed correct modification of each allele (S8 Fig). In addition, imaging of biallelic knock-in organoids confirmed the detection of both allele-specific reporters (Fig 4B). Next, we performed live-cell imaging of organoids and assessed the biallelic fluorescent output (green versus red) for single cells over time (Fig 4C), as a straightforward showcase how differential allele-specific modifications could be used to study allele-specific expression levels [18,19]. This proof of principle underscores the power of ITPN to create allele-specific readouts that, depending on the knock-in template, can be applied to address many biological questions, ranging from allele-specific expression patterns to differential biochemical properties between wild-type and mutant proteins.

[END]

[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001527

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/