An intersectional genetic approach for simultaneous cell type-specific labelling and gene knockout in the mouse Free

Single-cell RNA sequencing reveals cellular heterogeneity in organs, with especially high levels of heterogeneity in the brain (Darmanis et al., 2015; Vinsland and Linnarsson, 2022). In the field of neurobiology, one of the challenges is the precise labelling of specific cell types to accomplish gene manipulation in a cell type-specific manner in rodents, which currently depends on the mouse Cre-loxP system. However, undesirable expression of cell type-specific Cre, which could be due to transient expression during early development but shut down in adulthood, its natural expression in more than one cell type, kinetics of recombinases, sensitivity of the Cre reporter, and disruption in cis-regulatory elements by transgene insertion present a disadvantage when attempting to label heterogeneous cell subtypes for precise deletion of target genes in organs, especially the brain (Leshan et al., 2012; Taniguchi et al., 2011). An intersectional approach with the use of two site-specific recombinases (SSRs) can overcome this disadvantage in mice, thereby resulting in more precise genetic lineage tracing (He et al., 2017), specific cell-type targeting in the brain (He et al., 2016), and accurate gene knockout (Han et al., 2021) or activation (Awatramani et al., 2003). Transgenic mice and related adenoviruses have been developed for intersectional targeting with high specificity and performance (Madisen et al., 2015). The Flp-FRT and Dre-rox SSR systems have been used to target neural circuits with cell type and projection type specificity (Anastassiadis et al., 2009; Han et al., 2021; He et al., 2016; Karimova et al., 2018; Li et al., 2020; Liu et al., 2020; Madisen et al., 2015; Plummer et al., 2015; Sajgo et al., 2014). Although an intersectional approach for gene deletion and cell type-specific labelling has been achieved within defined brain regions using a retrograde virus or a combination of a Flp driver, a Flp-dependent-Cre virus, and a Cre-dependent fluorescence expression allele (Jung et al., 2019), mice with both cell type-specific labelling and gene knockout are not currently available.

To address this issue, we applied triple recombination comprising Cre-loxP SSR, Flp-FRT SSR and Dre-rox SSR to generate reporter-and-Dre-recombinase bicistronic knock-in mice and rox-flanked gene knockout mice. These engineered mice serve dual purposes of having both selective labelling of specific cell types and precise deletion of target genes within given cell types simultaneously. As a proof of principle, we labelled GABAergic neurons expressing neuropeptide Y (NPY), calretinin (calbindin 2) (CR) and cholecystokinin (CCK), using the intersectional crosses of corresponding Cre mice (NPY-Cre, CR-Cre and CCK-Cre) in conjunction with Flp mice expressing all GABAergic neurons (Dlx5/6-Flp). Next, using our engineered mice, we precisely deleted Rbfox3 (RNA binding fox-1 homolog 3) from each of the above three subtypes of GABAergic neurons using the Dre-rox system. RBFOX3 is a neuronal splicing regulator and regulates hippocampal circuit balance and function, in addition to adult neurogenesis and synaptogenesis (Huang et al., 2022; Lin et al., 2018, 2016b; Wang et al., 2015). We deleted target candidate genes in precise cell subtypes of corresponding conditional knockout (cKO) mice. This approach eliminates the problem of undesirable expression that occurs with cell type-specific Cre mice and provides a system for simultaneously labelling heterogeneous cell subtypes and deleting target genes within given cell subtypes in the mouse brain. The specificity of these cKO mice facilitate a more precise interpretation of gene expression in a cell type-specific manner. The use of cKO mice combined with other approaches would also provide information on causality between target gene candidates and their downstream functional readouts with cell type-specific resolution in addition to lineage tracing. In short, our triple SSR mice provide advantages for simultaneous cell type-specific labelling and genome engineering in vivo.

Cre-loxP, Flp-FRT and Dre-rox are three known systems that mediate SSR events in genomic DNA. An intersectional approach results in transgene expression that is more cell type-specific by combining different recombination systems that have distinct promoters and provide precise site-specific deletions. Therefore, we combined the Cre-loxP, Flp-FRT and Dre-rox systems to delete a target gene in a specific cell type of the mouse brain. First, we generated Dre knock-in mice with two STOP cassettes, with the first flanked by FRT and the second flanked by loxP, which precedes the tdTomato-P2A-DreO sequences (Fig. 1A, Fig. S1). Critical genomic sequences were validated by PCR and Sanger sequencing (Fig. 1B) and Southern blot (Fig. S1B,C). Next, we generated Rbfox3 (RNA binding fox-1 homolog 3) conditional knockout (cKO) mice with key exons flanked by rox sequences (Fig. 1C). Genotyping was employed to distinguish wild-type (WT), heterozygous (HET) and homozygous (Homo) cKO mice (Fig. 1D). The 5′ and 3′ rox sites were validated by Sanger sequencing (Fig. 1E,F). To further validate the occurrence of recombination and the presence of 5′rox and 3′rox sites on the same chromosome, we examined the rox recombination sites in DNA from the brains of Npy-Cre::Dlx5/6-Flp::Dre/+::rox/+ mice with 5′rox/F and 3′rox/R primers. PCR analysis demonstrated the correct rox recombination site in these mice (Fig. 1G). We also confirmed whether two STOP cassettes were removed under Cre and Flp expression in DNA from the brains of Npy-Cre::Dlx5/6-Flp::Dre/+::rox/+ mice with CAG/F and td/R primers using PCR analysis (Fig. 1H). Collectively, these findings validate the genomic structures of our Dre knock-in and Rbfox3^rox/rox mice.

Neuropeptide Y (NPY) is a 36-amino acid neuropeptide that is abundant in the brain and spinal cord (Danger et al., 1990; Wahlestedt et al., 1989; Wettstein et al., 1995). NPY is highly expressed in the basal ganglia, nucleus accumbens and amygdala, and moderately expressed in the hypothalamus, hippocampus, septal nuclei, cortex and periaqueductal grey matter (Danger et al., 1990; Wahlestedt et al., 1989; Wettstein et al., 1995). NPY is released from neurons, binds to NPY receptors, and is a molecular marker for distinct classes of GABAergic neurons and long projection cells in the brainstem (Silveira et al., 2020; Thorsell and Ehlers, 2006). In mice, the expression of Npy in the central nervous system (CNS) begins around embryonic day 11.5 (Yue et al., 2014). NPY-Cre mice are widely used for labelling NPY-expressing GABAergic neurons in the nervous system (Goswamee et al., 2021; Lee et al., 2022; Milstein et al., 2015; Tokarska and Silberberg, 2022). However, NPY-Cre mice showed undesirable expression of Cre recombinase in hippocampal dentate granule cells and excitatory neurons, as seen in NPY-Cre::Ai14 mice (Fig. 2B). Dlx5/6 is expressed in differentiating and migrating forebrain GABAergic neurons during embryonic development in Dlx5/6-Flp transgenic mice (Monory et al., 2006). The Dlx5/6 transgene was designed with the intergenic enhancer regions between the mouse Dlx5 and Dlx6 locus together with a β-globin basal promoter. Generation of NPY-Dre (NPY-Cre::Dlx5/6-Flp::Dre) mice solved this problem of undesirable expression as demonstrated in Fig. 2C. Furthermore, NPY-Dre mice showed high labelling efficiency and specificity of tdTomato⁺ cells to NPY-expressing GABAergic neurons (Fig. S2) and Cre⁺Flp⁺ cells (Fig. S3). Moreover, successful deletion of Rbfox3 in NPY-expressing GABAergic neurons was exhibited in the hippocampus (Fig. 2D-F) and cortex (Fig. S4A-C) of NPY-Dre-Rbfox3^rox/rox (NPY-Cre::Dlx5/6-Flp::Dre::Rbfox3^rox/rox) mice compared with their controls. Reduced density of tdTomato⁺ cells was observed in the hippocampus of NPY-Dre-Rbfox3^rox/rox mice compared with their controls (Fig. 2G). No cellular toxicity was observed in the hippocampus of NPY-Dre mice using terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assays (Fig. S5B).

Calretinin (CR; also known as calbindin 2, Calb2) is a calcium-binding protein that is highly expressed in the brain, particularly in adult cerebellum and frontal lobe (Yue et al., 2014). It is predominantly expressed in a subset of GABAergic neurons and transiently expressed in glutamatergic neurons during development (Camillo et al., 2014). Expression of Cr in the mouse CNS begins around embryonic day 11.5 (Yue et al., 2014). CR-Cre mice have undesirable expression of Cre recombinase in hippocampal dentate granule cells owing to transient expression in the early postmitotic stage of granule cell development (Brandt et al., 2003), which was demonstrated with immunostaining of the hippocampus from CR-Cre::Ai14 mice (Fig. 3B). Generation of CR-Dre (CR-Cre::Dlx5/6-Flp::Dre) mice successfully eliminated this problem of undesirable expression (Fig. 3C). Moreover, CR-Dre mice showed high labelling efficiency and specificity of tdTomato⁺ cells to CR-expressing GABAergic neurons (Fig. S6). Furthermore, successful deletion of Rbfox3 in CR-expressing GABAergic neurons was demonstrated in the hippocampus (Fig. 3D-F) and cortex (Fig. S7) of CR-Dre-Rbfox3^rox/rox (CR-Cre::Dlx5/6-Flp::Dre::Rbfox3^rox/rox) mice compared with their controls. CR is highly expressed in the excitatory mossy cells of ventral hippocampus (Fredes et al., 2021). Generation of CR-Dre (CR-Cre::Dlx5/6-Flp::Dre) mice successfully targeted CR-expressing GABAergic neurons of the ventral hippocampus (Fig. S8). In addition, successful deletion of Rbfox3 in CR-expressing GABAergic neurons of the ventral hippocampus was demonstrated in CR-Dre-Rbfox3^rox/rox (CR-Cre::Dlx5/6-Flp::Dre::Rbfox3^rox/rox) mice compared with their controls (Fig. S8B,C). Reduced density of tdTomato⁺ cells was observed in the hippocampus of CR-Dre-Rbfox3^rox/rox mice compared with their controls (Fig. 3G). No cellular toxicity was observed in the hippocampus of CR-Dre mice using TUNEL assays (Fig. S5C).

Cholecystokinin (CCK) is one of the most abundant neuropeptides in the CNS and is particularly concentrated in the cerebral cortex, hippocampus, basal ganglia, hypothalamus and periaqueductal grey matter (Raiteri et al., 1993). Cck is expressed in GABAergic and glutamatergic neurons (Calvigioni et al., 2017; Deng et al., 2010; Whissell et al., 2015). Expression of Cck expression in mouse CNS begins around embryonic day 11.5 (Yue et al., 2014). CCK-Cre mice exhibit undesirable expression in hippocampal dentate granule cells and excitatory neurons (Taniguchi et al., 2011), as also shown in CCK-Cre::Ai14 mice immunostained for CCK (Fig. 4B). Generation of CCK-Dre (CCK-Cre::Dlx5/6-Flp::Dre) mice eliminated undesirable expression in the dentate granule cells (Fig. 4C). Moreover, CCK-Dre mice showed high labelling efficiency and specificity of tdTomato⁺ cells to CCK-expressing GABAergic neurons (Fig. S9). In addition, successful deletion of Rbfox3 in CCK-expressing GABAergic neurons was demonstrated in the hippocampus (Fig. 4D-F) and cortex (Fig. S10) of CCK-Dre-Rbfox3^rox/rox (CCK-Cre::Dlx5/6-Flp::Dre::Rbfox3^rox/rox) mice compared with their controls. Increased density of tdTomato⁺ cells was observed in the hippocampus of CCK-Dre-Rbfox3^rox/rox mice compared with their controls (Fig. 4G). No cellular toxicity was observed in the hippocampus of CCK-Dre mice using TUNEL assays (Fig. S5D).

Currently, mice generated with an intersectional approach that allows for simultaneous labelling of cell type and deletion of a target gene within given cell types are unavailable commercially or from another research lab. To our knowledge, our lab is the first to use an intersectional approach for the generation of mice engineered with a target gene deleted in a cell type-specific manner in vivo. This approach eliminates the problem of undesirable expression and lack of precise cell type-specific markers in the brain that can occur with Cre mice. In addition, these mice also have had the target gene (Rbfox3) deleted within specific cell subtypes. Thus, the engineered mice described in this study provide innovative advantages for simultaneous cell type-specific labelling and genetic knockout in vivo.

An intersectional genetic approach was applied to cell labelling, ablation and lineage tracing, increasing our understanding of cell fate and heterogeneity. The enhanced precision of lineage tracing with an intersectional approach has been previously verified (He et al., 2017), and intersectional labelling has recently been applied in the field of neuroscience. Using an intersectional approach, cell fate over time was revealed in midbrain progenitor cells (Petese et al., 2022) and distinct subsets were identified in midbrain dopaminergic neurons (Tolve et al., 2021). An intersectional approach has also been used to examine subtypes of serotonin neurons in raphe nuclei (Ren et al., 2019) and subtype-specific organization of GABAergic neuronal inputs was revealed with a combination of intersectional labelling and rabies viral monosynaptic tracing (Yetman et al., 2019).

Second-order sensorimotor loops controlling vibrissa movements were identified with a combination of intersectional labelling, classical tracing and electrophysiological recordings (Bellavance et al., 2017). An intersectional approach has been used to identify a subpopulation of noradrenergic neurons (Plummer et al., 2015) as well as major classes and lineages of GABAergic neurons in cerebral cortex (He et al., 2016). An intersectional approach greatly improved the specificity and performance of neural sensors and effectors (Madisen et al., 2015). Labelling of heterogenous retinal ganglion cells was performed with an intersectional approach (Sajgo et al., 2014). Lineage tracing with an intersectional approach has also been applied to other fields, such as metabolism (Han et al., 2021). Our approach adds new value to this concept of intersectional genetic approach: simultaneous genetic knockout in additional to labelling of specific cell subtypes.

In the field of neurobiology, NPY-Cre, CR-Cre and CCK-Cre mice have been widely used for labelling NPY-, CR- and CCK-expressing GABAergic neurons in brain and manipulating their neuroactivity in vivo when investigating neural circuits. The use of an intersectional genetic approach to develop three engineered mice provides increased fidelity of cell type labelling, which is crucial in neurobiology research because of the heterogeneous cell types in brain. Furthermore, the ability to finely target cell types for gene deletion is another important issue in the field of neurobiology when there is a need to investigate the causal relationship between target genes and their corresponding functional readouts more definitively. Our intersectional Dre knock-in mice with Rbfox3^rox/rox mice can resolve the problematic issues resulting from the use of Cre mice lines and has the potential to provide an alternative platform for other Cre mice lines.

There is a fourth SSR system, the Vika-vox system, which has recently been shown to have a high efficacy and specificity in mice (Karimova et al., 2018). In the future, adding the Vika-vox system could also be employed to facilitate more precise cell type-specific labelling and genome engineering in our mouse lines. It will also be important to develop more cell type-specific Flp, Dre and Vika mice for more combinatory choices in intersectional genetic approaches because the current mouse library is composed predominantly of cell type-specific Cre mice. Recently, a suite of new Dre mice was developed, which expands the ability to perform intersectional genetic targeting (Han et al., 2021). In short, reliable and diverse reporter mice would facilitate the best use and application of our Dre-rox mice in the future.

It would also be of interest for future studies to pursue functional comparisons between Cre Rbfox3 conditional knockout mice, Dre Rbfox3 conditional knockout mice and Rbfox3 conventional knockout mice at the molecular, cellular, circuitry and behavioural levels. Such comparisons would strengthen the value of our Dre mice.

There are some limitations and practicalities to consider with our approach. For example, generating customized rox mice for specific candidate genes and breeding mice carrying homozygous rox sites in addition to Cre, Flp and Dre sequences requires extra efforts that were time consuming, labour intensive and expensive. Moreover, our approach cannot be applied to cellular subtypes for which an overlap of more than two genetic markers is required for specificity. Because there are more choices from Cre driver mouse lines than for Flp driver mouse lines, it might be possible to generate new Flp driver mouse lines to fulfil specific needs.

Combined genetic and viral methods have been used to target specific cell types in mouse brain (He et al., 2016). Our Dre::Rbfox3^rox/rox mice, which allow for cell type-specific labelling and genome editing, could be used in combination with adeno-associated virus (AAV)-carrying cell type-specific Cre and Flp to provide temporal and spatial resolution, respectively. Such application in the future would increase the feasibility, convenience and versatility of using our engineered mice.

Dre knock-in mice were generated by introducing a DreO cassette containing two STOP and tdTomato sequences into intron 1 of Gt(ROSA)26Sor (Fig. 1A) (Madisen et al., 2015). Specifically, the targeting vector containing sequences of left and right arms of ROSA26, two STOP, tdTomato, DreO, PGK-Neo and PGK-DTA was electroporated into embryonic stem cells (ESCs; C57BL/6N, JM8A3 strain) (Fig. S1). The original targeting vector was obtained from Addgene [Ai65(RCFL-tdT), #61577]. ESCs were cultured and positively selected by Neo and negatively selected by DTA. The target ESC clones were identified by Southern blot analysis (WT: 11.5 kb; targeted: 8.73 kb) (Fig. S1). One targeted ESC clone was microinjected into blastocysts (C57BL/6N strain) and transplanted into the uterus of pseudopregnant female mice. Three agouti chimeras were obtained; one was bred with B6 mice. Five out of 40 offspring (N1 generation) were agouti and their genotypes were verified with PCR using primers shown in Fig. 1A,B.

Rbfox3^rox/rox mice were generated by introducing two rox sites into the upstream of exon 7 and the downstream of exon 9 of Rbfox3, respectively with a CRISPR/Cas9 approach (Fig. 1C). Specifically, 5′ rox ssODN and 3′ rox ssODN and their corresponding sgRNAs and Cas9 were sequentially electroporated into zygotes. Edited zygotes (C57BL/6J strain) were transplanted into the uterus of pseudopregnant female mice. Three males and two females out of 19 F0 founders were identified and verified by PCR and TA cloning (Fig. 1E,F). Cis/trans testing was performed by their N1 generation after mating founders with B6 mice. The length of the rox flanking region is 4140 bp. Rbfox3 5′ sgRNA: 5′-AGTTTCTGGGTTCTAGAACG-3′. Rbfox3 3′ sgRNA: 5′-AATGACCCACCACTGCACGG-3′. Rbfox3 5′rox ssODN-SpeI (sense): 5′-TAACAGAGCCCAAGTGTCAAGTTTCTGGGTTCTAGAACTAGTTAACTTTAAATAATGCCAATTATTTAAAGTTAACGAGGTAAAGGGTTCCCCTCCTCTCAGAACCTGCAGACAAGGGTCCTATA-3′. Rbfox3 3′rox ssODN-SpeI (antisense): 5′-AGCCCATCTCGGAGGATAAGGGCAGCTGGGATGTGGTTCATCCTCCCTCCGTAACTTTAAATAATTGGCATTATTTAAAGTTAACTAGTTGCAGTGGTGGGTCATTGTGTTGTTTACGGATGGCA-3′. Male NPY-Cre/+::Dlx5/6-Flp/+::Dre/+::Rbfox3^rox/+ mice were mated to female NPY-Cre/+::Dlx5/6-Flp/+::Dre/+::Rbfox3^rox/+ mice to obtain NPY-Cre/+::Dlx5/6-Flp/+::Dre/+::Rbfox3^rox/rox mice and corresponding controls. Male CR-Cre/+::Dlx5/6-Flp/+::Dre/+::Rbfox3^rox/+ mice were mated to female CR-Cre/+::Dlx5/6-Flp/+::Dre/+::Rbfox3^rox/+ mice to obtain CR-Cre/+::Dlx5/6-Flp/+::Dre/+::Rbfox3^rox/rox mice and corresponding controls. Male CCK-Cre/+::Dlx5/6-Flp/+::Dre/+::Rbfox3^rox/+ mice were mated to female CCK-Cre/+::Dlx5/6-Flp/+::Dre/+::Rbfox3^rox/+ mice to obtain CCK-Cre/+::Dlx5/6-Flp/+::Dre/+::Rbfox3^rox/rox mice and corresponding controls. NPY-Cre mice (stock number 027851), and CR-Cre mice (stock number 010774) were purchased from The Jackson Laboratory. CCK-Cre mice (The Jackson Laboratory, stock number 012706) and Dlx5/6-Flp mice (The Jackson Laboratory, stock number 010815) were provided by the laboratory of C.-C.L. Mice were group-housed in ventilated cages, provided with food (PicoLab® Rodent Diet 20, 5053) and water ad libitum and maintained on a 12-h light/dark cycle (lights off at 20:00 h). The National Taiwan University College of Medicine and the College of Public Health Institutional Animal Care and Use Committee (IACUC) approved all procedures. Our animal facility has earned Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) accreditation. All experiments were performed in accordance with the approved guidelines.

PCR was run with SuperRed PCR Master Mix (TE-SR01, Biotools Co.) in a C1000 Touch Thermal Cycler (Bio-Rad). Detailed information regarding primer sequences, PCR product sizes and PCR cycling conditions are summarized in Table S1. To validate the 5′- and 3′-rox sites in Rbfox3^rox/rox mice, the amplified PCR product was cloned with the TOPO TA cloning Kit (Invitrogen, 450071).

Cellular toxicity was examined with the Elabscience®One-step TUNEL In Situ Apoptosis Kit (Elabscience, E-CK-A320), according to the manufacturer's manual. Briefly, brain tissue was fixed in 4% paraformaldehyde and 40 µm coronal frozen sections were collected. Brain sections were treated with DNase I for a positive control. DAPI was used as a marker of cell nuclei. Fluorescent images were acquired using a Zeiss LSM 880 confocal microscope (Carl Zeiss).

Mouse brains were perfused with 4% paraformaldehyde and sectioned coronally at a thickness of 40 µm from the hippocampus of postnatal day 49 mice. Details of immunofluorescence staining was conducted using methods previously described (Lin et al., 2016a; Wang et al., 2015). Information regarding primary and secondary antibodies are summarized in Table S2. In short, coronal sections were incubated with primary antibody at 4°C overnight, and then incubated with a corresponding secondary antibody and DAPI (1:50,000, Invitrogen, D-1306) for 2 h at room temperature. Images were acquired using a Zeiss LSM 880 confocal microscope (Carl Zeiss).

We thank the technical services provided by the ‘Transgenic Mouse Model Core Facility of the National Core Facility for Biopharmaceuticals, National Science and Technology Council, Taiwan’ and the ‘Gene Knockout Mouse Core Laboratory of the National Taiwan University Center of Genomic Medicine’ for generating Dre knock-in mice and Rbfox3^rox/rox mice. We also thank the staff of the imaging core and the biomedical resource core at the First Core Labs, National Taiwan University College of Medicine, for technical assistance.