Mutation of the Traj18 gene segment using TALENs to generate Natural Killer T cell deficient mice

Invariant Natural Killer T (iNKT) cells are a unique subset of T lymphocytes that have been implicated in both promoting and suppressing a multitude of immune responses. In mice, iNKT cells express T cell antigen receptors (TCRs) comprising a unique TCRα rearrangement between the Trav11 and Traj18 gene segments. When paired with certain Trbv TCRβ chains, these TCRs recognize lipid antigens presented by the major histocompatibility complex (MHC) class I-like molecule, CD1d. Until recently, the sole model of iNKT deficiency targeted the Jα18, which is absolutely required to form the TCR with the appropriate antigenic specificity. However, these mice were demonstrated to have a large reduction in TCR repertoire diversity, which could confound results arising from studies using these mice. Here, we have created a new NKT-deficient mouse strain using transcription activator-like effector nuclease (TALEN) technology to only disrupt the expression of Jα18, leaving the remaining Jα repertoire unperturbed. We confirm that these mice lack iNKT cells and do not respond to lipid antigen stimulation while the development of conventional T cells, regulatory T cells, and type Ib NKT cells is normal. This new mouse strain will serve as a new model of iNKT cell deficiency to facilitate our understanding of iNKT biology.


Invariant Natural Killer T (iNKT) cells are a unique subset of T lymphocytes that have been implicated in both promoting and suppressing a multitude of immune responses. In mice, iNKT cells express T cell antigen receptors (TCRs) comprising a unique TCRα rearrangement between the Trav11 and Traj18 gene
segments. When paired with certain Trbv TCRβ chains, these TCRs recognize lipid antigens presented by the major histocompatibility complex (MHC) class I-like molecule, CD1d. Until recently, the sole model of iNKT deficiency targeted the Jα18, which is absolutely required to form the TCR with the appropriate antigenic specificity. However, these mice were demonstrated to have a large reduction in TCR repertoire diversity, which could confound results arising from studies using these mice. Here, we have created a new NKT-deficient mouse strain using transcription activator-like effector nuclease (TALEN) technology to only disrupt the expression of Jα18, leaving the remaining Jα repertoire unperturbed.

We confirm that these mice lack iNKT cells and do not respond to lipid antigen stimulation while the development of conventional T cells, regulatory T cells, and type Ib NKT cells is normal. This new mouse strain will serve as a new model of iNKT cell deficiency to facilitate our understanding of iNKT biology.
Although the majority of α β TCR + T cells recognize peptide antigens presented by conventional polymorphic MHC I or MHC II molecules, a fraction of T cells deviates from this rule. Instead, these α β T cells recognize lipid or glycolipid antigens presented by members of the monomorphic molecules of the CD1 family, or they recognize microbial riboflavin precursor derivatives presented by the monomorphic MR1 molecules 1 . Of these, the most extensively studied lipid-reactive T cells are the natural killer T (NKT) cells, which detect a number of glycolipid antigens in association with CD1d.
Two broad classes of NKT cells have been defined on the basis of TCR expression and antigen reactivity 2 . Most studies of these cells focus on type I, or iNKT cells (for invariant NKT cells), which are the most prevalent NKT cells in mice 3 . iNKT cells express a TCR that is the product of a canonical rearrangement between the Trav11 (Vα 14) gene segment (Trav10 or Vα 24 in human) and the Traj18 (Jα 18) gene segment, with a CDR3α region invariant at the amino acid level 4,5 . This Vα 14 invariant chain is co-expressed with a limited set of Vβ chains, predominantly Trbv13-2 (Vβ 8.2), Trbv29 (Vβ 7) and Trbv1 (Vβ 2) in mice and Trbv25-1 (Vβ 11) in humans 4-7 . iNKT cells expressing these TCRs recognize several microbe-derived glycosphingolipid 8,9 and diacylglycerol antigens 10 , including the prototypical glycosphingolipid antigen α -galactosylceramide (α GC) 11,12 , and can be identified using CD1d tetramers loaded with this antigen 13,14 . Because of their unique ability to rapidly and potently secrete cytokines and influence downstream responses 15 , iNKT serve as an important link between the innate and adaptive immune systems and are often regarded as potential therapeutic targets. Studies have linked iNKT cell defects with increased susceptibility to a wide range of disease processes including autoimmunity 16 , cancer 17 , and even obesity 18 . Additionally, the potential for therapies involving iNKT cells is especially attractive since the use of α GC to activate iNKT cells has proven safe in humans [19][20][21] . Ever emerging roles for this unconventional subset of lymphocytes make the study of their development and regulation both relevant and significant.
In addition to iNKT cells, there exists other CD1d-reactive T cells, which do not express the invariant Vα 14-Jα 18 TCR. These cells were first described when Cardell and colleagues examined the TCR usage of T cell Scientific RepoRts | 6:27375 | DOI: 10.1038/srep27375 hybridomas generated using CD4 + T cells isolated from MHC II-deficient mice 22 . A subset of these cells were CD1d-reactive, but instead of expressing the canonical Vα 14-Jα 18 TCR, the vast majority surprisingly expressed a heterogeneous TCR repertoire 22 . These CD1d-reactive cells have been named type II NKT cells or vNKT (for  variant NKT cells 23 ) because they do not express the canonical Vα 14-Jα 18 TCR, do not react with α GC, and tend  to express a variable TCR repertoire 2 . In order to study the functions of NKT cells directly in vivo, various genetically engineered strains of mice that lack NKT cells have been produced. Two approaches have been taken. In the first one, the Jα 18 gene segment was deleted from the mouse genome through homologous gene recombination 11 . Because the amino acids coded by the Jα 18 segments in the final TCRα chain of the iNKT TCR are absolutely essential to the recognition of CD1d 24 , no iNKT cells develop in absence of Jα 18, as no other Jα gene segment appears to be able to compensate 11 . However, because most type II NKT cells do not use the Jα 18 gene segment as part of their TCRs, this population is expected to still be intact in Jα 18 deficient mice.
The second approach aimed at disrupting the expression of CD1d molecules, which is required for NKT cell development [25][26][27] . However, in contrast to the Jα 18 null mouse, the CD1d deficient mice would be expected to lack both iNKT cells and type II NKT cells. Therefore, by comparing the phenotypes obtained among Jα 18 deficient, CD1d deficient, and wildtype (WT) mice in various immunological models, one should be able to identify potential roles of either iNKT or type II NKT cells. Such a "trick" has now been used in many studies (see ref. 28 for review) and due to the absence of markers that distinguish type II NKT cells, this technique remains the primary means to study the potential function and influence of type II NKT cells on the immune system.
Recently, we reported that T cells derived from the original Jα 18 null mouse (Tcrα -J tm1Tgi , called Jα 18(neo) here) model of iNKT cell deficiency failed to express TCRs using Traj gene segments that are located 5′ of the Traj18 gene in the genome 29 . Since non-productive rearrangements were similarly affected in these mice, we speculated that this may be due to a genetic event, possibly due to adverse consequences that the PGK-neomycin selection cassette used to generate this line might have had on transcription. The replacement of the Jα 18 gene with PGK-neomycin in the reverse orientation with respect to the Jα region transcription could introduce a constitutively open chromatin configuration while the PGK promoter could serve as competition for transcription factors 30 . As a consequence, we estimated that TCRα diversity in these mice was reduced by about 60%, raising the possibility that some of the observations previously generated with these mice might be due to an incomplete TCR repertoire rather than the absence of iNKT cells.
Here, we report the generation of a new Traj18 deficient mouse using transcription activator-like effector nuclease (TALEN) technology. Our new mouse model eliminates production and expression of the requisite iNKT cell Vα 14-Jα 18 junction upon recombination but preserves Jα gene usage for all other Traj genes to allow for proper T cell development. This new mouse provides a model to unambiguously understand the contributions of iNKT cells in in vivo models.

TALEN design and Traj18 mutant mouse generation. Two TALENs (called TALN 1 and TALN 2)
were designed to target the Traj18 gene segment (Fig. 1A). The TALENs were able to recognize and cleave the Traj18 target sequence using an in vitro system, as described previously 31 . Briefly, a reporter plasmid that constitutively encodes RFP followed by the target sequence (Traj18) and out of frame GFP was constructed. When nuclease activity occurs, the double stranded break is repaired via error prone non-homologous end joining, creating genomic insertions or deletions (indels). Many of these indels result in frameshift mutations, putting the GFP in frame, and thus permitting its functional expression. This method allows for rapid assessment of TALEN directed mutation activity. We transfected HEK 293 cells with plasmids encoding either TALN 1 or TALN 2 and the reporter encoding the target sequences for either TALN 1 or TALN 2. Irrelevant target DNA was not cleaved, as shown by absence of GFP expression when TALN1 was co-transfected with the surrogate reporter for TALN2 and vice versa, while both TALN 1 and TALN 2 were able to cleave their targeted sequence (Fig. 1B).
RNA for each of the TALEN pair were microinjected into pure C57BL/6 zygotes. Pups derived from the injected zygotes were genotyped by direct sequencing of genomic DNA. All mice showed modifications of the Traj18 gene sequence (not shown). Because only one Traj18 allele might have been affected and/or both alleles might have been modified differently, we cloned PCR products amplifying the Traj18 gene and sequenced plasmids individually. The results are summarized in Fig. 2. We selected one founder (J14) with a 10 base pair (bp) deletion in the region of the gene contributing to the CDR3. We then established the line that we refer to as Jα 18(-10) here by crossing founder J14 with C57BL/6 females. Heterozygous offspring from this cross were then bred together to generate the homozygous Jα 18(-10) line. The segment deleted includes the germline-encoded Jα 18 CDR3 residues that are identical among mouse, rat, and human, and critical for TCR recognition and binding of the CD1d-α GC complex. Finally, because TALENs have the potential for cleaving the DNA at off-target sites, we also amplified and sequenced the top 3 potential off-target sites. All sequences were identical to the wild type genome (data not shown).
The Traj repertoire diversity is preserved in Jα18(-10) mice. Given the unexpected effect on the TCRα repertoire in the original Jα 18(neo) mice, we wanted to examine the TCRα diversity in the Jα 18(-10) mice. Pre-selection double positive thymocytes (DPs; CD4 + CD8 + CD69 − ) from the thymi of C57BL/6, Jα 18(neo), and Jα 18(-10) were sorted using flow cytometry. After generation of cDNA from these cells, we amplified Vα 14 (Trav11) TCR rearrangements through the use of a specific forward primer for Vα 14 and a reverse primer specific for the gene encoding the TCRα -chain constant region (Cα ; Trac). PCR products were subjected to Next-generation sequencing using the Ion Torrent platform. Focusing on productive in-frame rearrangements, we found that the frequency usage of Jα gene segments in Jα 18(-10) preselection DPs was comparable to C57BL/6 Scientific RepoRts | 6:27375 | DOI: 10.1038/srep27375 while the Jα 18(neo) lacked TCR transcripts using Jα segments upstream of Jα 18, confirming our previous data 14 (Fig. 3).
Jα 18 continues to be used in the Jα 18(-10) mice, albeit at a lower percentage than in C57BL/6 WT mice (Fig. 3). Since the Jα 18(-10) mouse has an intact recombination signal sequence (RSS) for the Jα 18 gene, the partial gene can still undergo recombination and presumably generate productive rearrangements. In the C57BL/6 preselected repertoire, the CDR3α length of Vα 14-Jα 18 rearrangements is distributed between 14 and 17 amino acids, with the majority of CDR3α transcripts consisting of 15 amino acids, the length of the canonical NKT CDR3α (CVVGDRGSALGRLHF). However, in the Jα 18(-10) mice, the CDR3 length distribution of Vα 14-Jα 18 rearrangements spanned between 9 and 13 residues. This variation was not due to a differential activity of the enzyme terminal deoxynucleotidyl transferase (TdT), which adds non-template nucleotides, but was due to the targeted 10bp deletion in Jα 18 (data not shown). This equates a three to four amino acid loss in the TCRα product, represented by the shift in CDR3α size (Fig. 3).

Vα14-Jα18 iNKT cells are absent in the Jα18(-10) mice but thymic development of conventional αβ T cells is normal.
To assess if the deletion of 10 bp in Jα 18 would affect the development of T cells, we characterized the various thymic subsets using flow cytometry. The proportions of double positive, double negative, CD4, and CD8 single positive thymocytes were identical between C57BL/6 and Jα 18(-10) mice. We assessed the percentage of mature T cells, as measured by the upregulation of TCR and the downregulation of CD24 (HSA), and observed no differences (Fig. 4A). Additionally, total thymic lymphocyte cellularity was similar between C57BL/6 and Jα 18(-10) mice (Fig. 4B).
We wondered if thymic regulatory cell (Treg) development, which has more stringent requirements for TCR specificity and self reactivity than bulk conventional α β thymocytes (reviewed in 32 ) was perturbed. Assessment by percentage and number revealed that Treg development was not affected by the 10 bp Jα 18 deletion (Fig. 4C). Surprisingly, the Jα 18(neo) had a greater percentage of regulatory T cells in the CD4 pool, which could be a byproduct of Jα repertoire skewing (Fig. 4C).
Next, we assessed the proportion and iNKT cell numbers in the thymus, spleen, and liver by flow cytometry using PBS57 (an α GC analog) -loaded CD1d tetramers that specifically stain iNKT cells 33 . Over one million events were collected per sample in all organs to ensure that the data generated were representative. As seen in Fig. 3D, the proportion and numbers of iNKT cells in the thymus, spleen, or liver of the Jα 18(-10) mice were close to background levels, similar to what is found with the original Jα 18 (neo) mice (Fig. 4D,E). Type Ib NKT cells in the Jα18(-10) mice. In addition to iNKT cells that use the canonical Vα 14-Jα 18 rearrangement, another minor population of α GC-reactive NKT cells was recently described 34 . These cells express a canonical Vα 10-Jα 50 TCRα chain and were named type Ib NKT cells 34 . They were originally identified in the Jα 18(neo) mice, which minimally express TCRα rearrangements that involve the Jα 50 segment (see Fig. 3 and 29 ). Because the new Jα 18(-10) mice had restored normal Traj frequency distribution, we wondered if we would detect a larger Vα 10Jα 50 type Ib NKT population in these mice. Although type Ib NKT cells recognize α -GC/CD1d complexes, they show a preference for the recognition of α -glucosylceramide (α -GlcCer) presented by CD1d 34 . Therefore, we stained the thymocytes from C57BL/6 and Jα 18(-10) mice with both of these tetramers. A minor population, representing about 0.01% of total thymocytes could be detected using both tetramers but was absent from CD1d-deficient mice (Fig. 5A). To ensure the legitimacy of this population, we enriched the cells using α -GlcCer/CD1d tetramers and magnetic bead cell isolation (Fig. 5B). Using this enrichment, we could detect a small but genuine population of type Ib NKT cells in the thymus of Jα 18(-10) mice. The cells included CD4 + and CD4 − subsets, were CD44 high and CD69 int (Fig. 5C), as previously reported 34 . We also examined the expression of the three major transcription factors (PLZF, T-bet and RORγ t) that were recently used to define functionally distinct iNKT cell subsets 35 . NKT1 cells are defined as T-bet hi , PLZF lo RORγ t lo , NKT2 cells are T-bet lo , PLZF hi and RORγ t lo , while NKT17 cells are T-bet lo , PLZF int and RORγ t hi . As seen in Fig. 4D, the proportion of NKT2 cells was lower and the proportion of NKT17 cells was higher in NKT type Ib as compared to Vα 14Jα 18 type I iNKT cells (Fig. 5D). Altogether, these results confirm a population of Jα 18 independent NKT cells that are neither classical type I nor type II and are probably not limited in frequency solely due to a reduced Jα repertoire.
Jα18(-10) mice do not respond to αGC challenge in vivo. Upon stimulation with α GC, iNKT cells are selectively activated and rapidly secrete large quantities of IL-4 and IFN-γ cytokines 14,36 . As seen in Fig. 5, upon α GC injection into C57BL/6 mice, large quantities of IL-4 and some IFN-γ were detected in the serum two hours post-injection. At 18 hours post-injection, large quantities of IFN-γ , mostly due to the transactivation of NK cells by activated iNKT cells 36,37 , were detected in the serum of C57BL/6 mice. In contrast, we did not detect any of these cytokines when the experiment was performed with Jα 18(-10) mice (Fig. 6). It is likely that the cytokine quantities released by activated type Ib NKT cells were too low to be captured by our assay. Altogether, these results further demonstrate the absence of functionality associated with the iNKT cell population in the Jα 18(-10) mice.

Discussion
We have previously reported that the original Jα 18 null mouse had defects in the TCRα repertoire beyond the absence of the Jα 18 gene segment 29 . As a consequence, we estimated that a large portion of TCR diversity was lost in these animals. Since the creation of the Jα 18(neo) in 1997, these mice have been the sole model of type I NKT cell deficiency. To avoid the confounding and unintended results due to reduced TCR diversity, every study in the twenty-year span that used these mice as a model of NKT cell deficiency or to differentiate between type I versus type II NKT cell consequences would benefit from a reassessment. We have created a mouse that is deficient only for type I NKT cells that use the invariant Vα 14-Jα 18 TCR chain. This new Jα 18 null mouse has considerable advantages over the original line, preserving the Jα diversity of the WT animal while eliminating expression of the invariant NKT TCR. Therefore, we expect not only the full restoration of T cell repertoire diversity but also the restoration of other unique T cell populations with limited and nearly exclusive Traj usage that might have been lost in the Jα 18(neo) mice. For example, TCRs expressed by mucosal-associated invariant T (MAIT) cells are composed of rearrangements involving Trav1 and primarily Traj33 in mice 38 . Because Traj33 is located 5′ of Traj18 in the Traj locus, it is likely that the overall population size of MAIT cells is affected in the Jα 18(neo) mice. The restoration of Jα s proximal to Jα 18 presumably allows for normal development of this population in the Jα 18(-10) mice. Interestingly, frequencies of Jα 18 independent type Ib NKT cells were not appreciably increased despite more Traj50 representation in the preselected repertoire. Thus, other factors must govern the number of developing type Ib NKT cells such as limited thymic niche, ligand availability or genetics since these cells are more abundant in the BALB/c than C57BL/6 background 34 .
Recently, Chandra et al. generated and described another Jα 18 deficient using bacterial artificial chromosome and cre/lox technology to remove both Traj18 and the neomycin cassette 39 . We, instead, elected to use targeted genome engineering and created our line with TALEN technology. Using this method, we prevented expression of the iNKT TCR without eliminating Jα 18 expression. In the Jα 18(-10) mice, the preselected T cells using the Jα 18 gene segment had CDR3α lengths on average 3-4 amino acids shorter than what is found in wild-type mice. Presumably, some of these TCRs can be positively selected and expressed. Additionally, the Jα 18 RSS is undisturbed, allowing for normal recombination events. Nonetheless, by disrupting the size of the CDR3α loops that Order of gene segments along horizontal axes (left to right) is similar to their 5′ to 3′ organization in the mouse genome. Rearrangements for were amplified by PCR with a V-specific primer and a C-specific reverse primer (above plots), followed by high-throughput sequencing with the Ion Torrent platform. Sequence analysis was performed with inhouse software, and gene identity was assigned on the basis of sequence alignment with published sequences (International ImMunoGeneTics Information System). can be generated, we completely abrogated the development of iNKT cells, in agreement with the importance that this loop has in interacting with the antigen-CD1d complex 40 .
We have shown that the Jα 18 (-10) mouse is a precise model of iNKT deficiency while maintaining the development of other CD1d-restricted T cells, which do not require the Vα 14-Jα 18 rearrangement. Unlike the Jα 18(neo) mice that have reduced TCR repertoire diversity, the TCR repertoire in Jα 18(-10) mice appears similar to WT mice. We believe this novel mouse will serve as a new standard model for iNKT cell deficiency and eliminate the confounding factors that have surfaced with the original mouse model.  11,26 . C57BL/6 were purchased from Jackson Laboratories. All mice were used between 6 to 12 weeks and were age matched for each experiment. All mice were raised in a specific pathogen-free environment at the Biology Resource Center in National Jewish Health or the University of Colorado Anschutz Medical Campus Vivarium. All Experiments were approved by the NJH and the University of Colorado Institutional Animal Care and Use Committees and performed in accordance with their guidelines.
Lymphocyte Isolation. Single cell suspensions were prepared from the thymus, spleen, and liver by manual disruption using syringe plunger. The liver was perfused with PBS then cut into small pieces, disrupted with a syringe plunger, and liver lymphocytes were isolated by centrifugation using a 33% (vol/vol) Percoll gradient (GE Healthcare).
High-Throughput Sequencing. CD69 − CD4 + CD8 + pre-selection double positive thymocytes were sorted by flow cytometry and washed twice in ice-cold PBS. Total RNA was extracted using the RNeasy Mini Kit (Qiagen). cDNA was made using SuperScript III Reverse Transcriptase (Invitrogen). The Vα 14-Cα region was amplified using specific Vα 14 (5′ -TACAGTGTGACCCCCGACAAC-3′ ) and Cα (5′ -GAGGGTGCTGTCCTGAGACCGAG-3′ ) primers with required specific Ion Torrent tags. Purified PCR products were sent for high-throughput sequencing using the Ion Torrent platform. Sequence analysis was done with in-house software, and gene identity was assigned on the basis of sequence alignment with published sequences (International ImMunoGeneTics Information System). Cells were analyzed on a BD LSRFortessa (BD Biosciences) and data were processed with FlowJo software (TreeStar). Enrichment of CD1d reactive thymocytes. Thymocytes were enriched for CD1d-α -galactosylceramide reactive cells by incubating thymocyte cell suspensions with PE conjugated CD1d-α -glucosylceramide for 45 minutes at 4 °C, then incubated with anti-PE magnetic microbeads (Miltenyi Biotec) for 15 minutes at 4 °C, followed by separation by using an autoMACS Pro Separator (Miltenyi Biotec) according to manufacturer's instructions. Cells were then later surface stained, fix/pearmeabalized, and then intracellular stained for flow cytometry. Fixable Viability Dye from Affymetrix eBioscience was used to exclude dead cells in enrichment experiments.
In vivo Cytokine Response Quantification. Mice were administered vehicle or 2 μ g of α GC (Alexis Biochemicals) by intravenous injection. Blood was collected at 2 hours and 18 hours post injection and serum was isolated using Z-gel microtubes (Sarstedt). Serum cytokine levels were measured by ELISA using Mouse Th1/ Th2 ELISA Ready-SET-Go! (Affymetrix eBioscience) according to manufacturer's instructions. Statistical analysis. Prism software (GraphPad) was used for statistical analysis. Unpaired, two-tailed Student's t tests were used for generation of p-values. Figure 6. Jα18(-10) mice do not respond to αGC challenge in vivo. C57BL/6 and Jα 18(-10) mice received 2 μ g of α GC intravenously. Serum was collected at 2 and 18 hours after injection. IL-4 and IFN-γ was quantified by enzyme-linked immunosorbent assay. Data shown represent mean ± SEM for each group (minimum of 3 mice per group). * * P < 0.01; * * * P < 0.001.