In order to recognize and eliminate the many different pathogens that the immune system can encounter, it generates an enormous repertoire of different unique T-cell receptors and B-cell immunoglobulins. The main contributor to this diversity is the process of V(D)J recombination, in which first different V, (D) and J genes are combined, then more diversity is created at the junction sites between the V, (D) and J genes. Recombination-activating gene 1 (RAG1) and 2 (RAG2) are crucial both for the recognition and cleavage of the RSS flanking the V, D or J genes and to form the hairpins that allow the broken DNA ends to be ligated by the non-homologous end joining pathway (NHEJ). Mutations in RAG1 and RAG2 are associated with a broad spectrum of distinct clinical and immunological phenotypes, including T- B- severe combined immune deficiency (SCID), Omenn syndrome (OS), atypical/leaky SCID (LS), γδ SCID and combined immunodeficiency with granuloma and/or autoimmunity (CID-G/AI). Previous to this work, in vitro assays and structural modeling of the RAG complex have given some insights into why some mutations cause a more severe phenotype than others. In particular, mutations that allow for more residual RAG activity result in phenotypes such as leaky SCID or CID-G/AI. However, phenotypic variability remains even between patients with mutations that affect the same region of RAG1, showing the same residual RAG1 activity using in vitro models. How these mutations cause these different phenotypes remains an open question. The main aim of this thesis was to generate novel in vitro and in vivo models of RAG1 deficiency to study the mechanisms underlying the different RAG phenotypes. These models will also be useful to test novel treatment options for patients with mutations in RAG1. A promising novel treatment is gene correction, so improving the genome-editing strategy used to generate models of RAG1 deficiency should also contribute to the development of a gene therapy that corrects patient stem cells. Genome editing using CRISPR/Cas9 In chapter 2 we review the most important genome-editing strategies for primary immunodeficiencies. Currently, the mainstay of treatment for SCID or other severe primary immunodeficiencies is allogeneic hematopoietic stem cell transplantation, and although outcomes have improved over the years, this remains a high-risk procedure when HLA-matched donors are lacking. For this reason, correcting the patient’s own hematopoietic stem cells with gene therapy would offer an attractive alternative. Gene therapies currently being used in clinical settings insert a functional copy of the entire gene by means of a viral vector. Even though safer vectors continue to be developed, this strategy risks integration within oncogenes. In the case of RAG deficiency, gene addition is even more challenging, because RAG expression is so tightly regulated, and continuous expression can be harmful. A promising alternative is the use of endonucleases such as ZFNs, TALENs and CRISPR/Cas9 to introduce a double-stranded break in the DNA and thus induce homology-directed repair. With these genome-editing tools a correct copy can be inserted in a precisely targeted “safe harbor.” They can also be used to correct pathogenic mutations in situ and to develop cellular or animal models needed to study the pathogenic effects of specific genetic defects found in immunodeficient patients. The chapter focuses on CRISPR/Cas9 because of its efficacy and versatility. Models created from iPSCs Chapter 3 describes how we generated and characterized induced pluripotent stem cells (iPSCs) of one Omenn patient and two SCID patients. The Zúñiga- Pflücker lab used their in vitro assay to differentiate these iPSCs and the iPSCs of healthy donors into T-cells. The SCID and Omenn iPSCs showed a similar block in T-cell development. Subsequently, we compared the TCR repertoire of the precursor T-cells and found that cells derived from SCID and Omenn patients showed a restricted repertoire with reduced number of7 rearrangements and skewed usage of V and J genes. In addition, T-cell receptor excision circle (TREC) levels, a measure of recombination activity also used for newborn screening, were reduced. The in vitro differentiation of iPSCs from RAG1 patients into T-cells described in chapter 3 will be useful to test gene-correction with CRISPR/Cas9. I was able to target and introduce a double stranded DNA break in the RAG1 locus in iPSCs, but did not yet succeed in true gene correction due to the challenges of first getting enough CRISPR/Cas9 and repair DNA template (single stranded oligonucleotide) into the cells, and subsequently getting enough homologous recombination to correct the mutation. As yet no lab has succeeded in CRISPR/ Cas9 mediated gene correction for RAG1 deficiency, though such corrections have been reported for some of the other SCID causing mutations. Creating a Rag1 mouse model using CRISPR/Cas9 Although they are useful to test true gene correction and the intrinsic effects of a mutation on the T-cell differentiation potential, in vitro models provide an artificial setting and cannot simulate the complexity of the full immune system with all its different components. In addition, the effect of environmental factors cannot be tested. To control for environmental factors in a realistic setting we need in vivo models. For this reason, I generated several Rag1 mouse models. For the study reported in chapter 4 I showed that CRISPR/Cas9 can be used to generate Rag1 mouse models in a single step, by directly injecting the Cas9 and Rag1 specific gRNA into the zygote. Residue 838 was targeted with high efficiency, as none of the mice showed a wild type sequence. This way, we were able to evaluate the effects of in-frame deletions in residues 832-877 of Rag1, a region of unclear function that does not contain any catalytic residues nor is involved in zinc binding. It was found that even 1 amino acid deletions in this region resulted in a complete knock-out phenotype with a complete lymphocyte developmental block in thymus at DN3 and bone marrow at the pro-B-cell stage. In addition, a H836Q missense mutation had no effect at all on RAG function. This provided insights into the RAG1 structure and function. In addition, this study offered proof of principle that a similar strategy can be applied to develop novel knockin mouse models with specific patient mutations. Mouse models of Rag1 mutations found in patients with CID-G/AI The underlying mechanisms causing the autoimmunity in patients with CID-G/AI are unclear. Previously, it was found that peripheral blood T- and B-cells of RAG patients with CID-G/AI have an abnormal T-cell receptor and B-cell immunoglobulin repertoire. However, it was unknown if these changes were already present in the primary repertoire in thymus and bone marrow, or may simply be due to antigenic stimulation in the periphery. Therefore, for chapter 5 I used CRISPR/Cas9 to generate three mouse models with missense mutations equivalent to those found in patients with CID-G/AI. Immunological characterization showed partial development of T and B lymphocytes, preserved serum immunoglobulin, partial immunity and presence of autoantibodies, thereby recapitulating the phenotype seen in patients with CID-G/AI. By using high-throughput sequencing, we identified marked skewing of IghV and TrbV gene usage in early progenitors, with a bias for productive Igh and Trb rearrangements after selection occurred and increased apoptosis of B-cell progenitors. Rearrangement of light chain was impaired, and polyreactive immunoglobulin M antibodies were detected. This study provides novel insights into how hypomorphic Rag1 mutations alter the primary repertoire of T- and B-cells, setting the stage for the immune dysregulation frequently seen in patients. Conclusion This thesis provides novel insights into the phenotypic spectrum of RAG1 deficiency, with a focus on the CID-G/AI phenotype. The mouse models I generated using CRISPR/Cas9 mediated genome-editing can be used for further investigations of autoimmunity in RAG1, as evidenced by several ongoing studies described in chapter 6. Mouse models such as these can also be used to develop novel treatment strategies and to optimize existing ones. For instance, hematopoietic stem cell transplantations for RAG patients with CID-G/AI are complicated by the requirement for conditioning regimens to get rid of the patient’s own (dysfunctional) T- and B-cells. Toxicity related to conditioning regimens based on the use of chemotherapy and/or irradiation remains a significant problem and novel conditioning regimens or other treatment strategies are needed. Possible solutions can be safely tried on mouse models. The in vitro T-cell differentiation from patient iPSCs discussed in chapter 3 can be used to further develop genome-editing correction strategies and offers a pre-clinical platform to prove true gene correction in human stem cells.