The Role of PTPN22 Risk Variant in the Development of Autoimmunity: Finding Common Ground between Mouse and Human

Genetic variation plays a pivotal role in the development of autoimmune disease. In the past decade the complexity of the genetic factors that influence the development of autoimmunity has become increasingly clear through the explosion of genetic information provided by both genome wide-association studies and whole-genome/exome sequencing. However, extending our knowledge from the genetic variations associated with disease to the pathogenic mechanisms that trigger, promote, or sustain autoimmunity continues to be a challenge. A key goal of such work is to understand how genetic risk variants contribute to development of disease and, more importantly, to determine whether these variants help to identify a set of common pathways that can be targeted for diagnostic and therapeutic purposes.

Several tools are available that help us address the functional consequences of a genetic variant: use of cell lines into which the genetic variant of interest can be introduced, murine models that either lack the gene linked to autoimmunity or contain the variant gene via transgene or KI strategies, and studies of human cells derived from healthy individuals and from individuals with disease who carry risk variants of interest. Each of these approaches has added to our understanding of the potential impact of risk or protective alleles on the human immune response, and these tools are being increasingly applied to genes associated with autoimmunity. However, it remains crucial to recognize that each approach has unique strengths: cell lines allow for extensive biochemical analysis, murine models allow us to dissect how variants impact the immune system’s development and identify mechanisms that may cause autoimmunity, and human tissue/blood analyses allow us to assess whether a genetic variant in humans may impact intrinsic immune cell function and/or modulate such responses following environmental exposures. Each approach also has specific limitations that may lead to conflicting results that must be interpreted within this context. In this review, we focus on studies of one genetic variant, PTPN22 C1858T, and discuss how the combination of these tools can allow us to develop a better understanding of how genetic variants contribute to autoimmunity.

In 2004, a single nucleotide polymorphism (SNP) in the protein tyrosine phosphatase N22 (PTPN22), 1858C > T (rs2476601), was shown to be associated with both type 1 diabetes (T1D) (1) and rheumatoid arthritis (2). Subsequently, these findings were confirmed in T1D and rheumatoid arthritis, and the variant has been linked to additional diseases, including systemic lupus erythematosus, Graves’ disease, and myasthenia gravis (3–5). This individual SNP is one of the most strongly associated autoimmune risk variants, after HLA class II, with odds ratios ranging from 1.9 to 1.3 (6). PTPN22 encodes the protein commonly referred to as lymphocyte tyrosine phosphatase (Lyp), and the 1858C to T polymorphism results in a single amino acid change from arginine (R) at position 620 to tryptophan (W). This amino acid change has become the focus of studies to determine how the PTPN22 1858T variant contributes to autoimmunity.

Lyp is part of the PEST group of nonreceptor classical class I PTPs and is expressed in all hematopoietic cells; its function has been reviewed extensively elsewhere (7). It is clear that Lyp is a negative regulator of TCR signal transduction via interactions with phosphorylated Lck (Y394), Fyn (Y427), and Zap-70, as well as phospho-sites on TCRζ, CD3ε, Vav, and Vcp (8, 9). Lyp’s function is thought to be augmented by its interaction with the C-terminal Src kinase (Csk), mediated by binding of the SH3 domain of Csk to the proline-rich region 1 of Lyp. The interaction between Csk and Lyp results in enhanced inhibitory tuning of Src family kinases and dampens TCR signal transduction (10). However, the negative outcome of this interaction has been challenged; additional studies suggest that the Lyp/Csk complex can function to limit Lyp activity via promoting phosphorylation of an inhibitory residue on Lyp (11) and/or by altering its localization within cell membrane lipid rafts (12). More recently, TRAF3 also was shown to bind the proline-rich region 1 of Lyp in myeloid cells and impact its turnover (13). Finally, additional binding partners of Lyp have been identified, including the adaptor molecule Grb2, the E3 ligase c-Cbl (14), and other signaling effectors (15) (R. James, X. Dai, and D.J. Rawlings, unpublished observations), implying that alterations in Lyp-dependent protein–protein interactions might impact signaling through additional events downstream of TCR, BCR, and other immune receptors.

Jurkat T cells lines were used in initial studies of Lyp versus the Lyp risk variant on TCR signaling. Cotransfection of the variant with Csk resulted in reduced interaction with Csk (compared with wild-type [WT] Lyp) (1) and an increase in TCR responses (16). In contrast, when Lyp620W was transfected into Jurkat cells, without the addition of Csk, a blunting of the TCR signal was observed (17). These differences, with respect to the impact on TCR signaling, led to persistent questions as to whether the variant is a gain-of-function versus a loss-of-function allele. These signaling analyses are clearly impacted by the relative expression levels of Lyp and Csk, which may expose different mechanisms by which Lyp regulates TCR signaling. In the setting of limiting Csk, the localization of Lyp to the cell membrane may be enhanced, resulting in increased inhibitory activity; in cell lines in which Lyp and Csk are expressed at a similar level, the loss of interaction between the variant and Csk may emphasize its loss of function in the inhibition of TCR signaling.

The study of unmanipulated human cells benefits from the ability to determine which functional findings from cell lines are likely most relevant, because the function of cells taken directly from the body reflect the expression and interaction of proteins found in vivo. However, such studies are limited by the numbers of cells available, as well as both the genetic and environmental variability within the multiple subjects that must be studied. Despite these limitations, several significant observations were made related to the role of Lyp620W and immune function. Initial studies by Vang et al. (17) demonstrated a decrease in IL-2 production after TCR stimulation in T1D subjects carrying the PTPN22 1858T variant, a finding that was confirmed in a larger cohort (18). Consistent with these findings, studies in healthy carriers of the risk variant identified blunted Ca²⁺ flux, IL-2, and IL-10 production after stimulation with anti-CD3, a finding that was most pronounced in the memory T cell compartment (19). Analysis of individuals with the variant allele and ANCA vasculitis also showed a decrease in IL-10 production (20). Further evidence of the downstream effect of the PTPN22 1858T variant on T cell maturation includes the observation that the CD4 memory T cell compartment is expanded (19) and that there is an increase in Th1 cells (21).

Lyp is expressed in B cells and influences BCR signaling and cell survival. An increase in Lyp expression is observed in chronic lymphoid leukemia and correlates with blunted BCR signaling via Syk and PLCγ, enhanced Akt activity and prosurvival signals, and an increased ability to escape BCR-mediated cell death (22). Lyp620W expression in B cells also was shown to alter BCR signaling, B cell maturation, and tolerance in studies of normal human B cells. Menard et al. (23) found an increase in the frequency of autoreactive naive mature B cells in healthy carriers of Lyp620W. Moreover, Habib et al. (24) demonstrated an expansion of the transitional B cell compartment and increased numbers of anergic B cells, findings that correlated with a blunting of BCR signaling and resistance to BCR-mediated cell death among immature B cells and are anticipated to contribute to a loss of B cell tolerance. These studies indicate that Lyp620W may directly contribute to the development of autoantibodies and autoimmunity through its impact on B cell development.

Recent work also implicated Lyp in modulating innate signals in myeloid cells, specifically through its influence on TLR signaling and subsequent production of type 1 IFN. Wang et al. (13) described a failure of the Lyp variant to promote the ubiquitination of TRAF3 that is required for downstream IFN production. Primary murine myeloid cells expressing the human Lyp variant (and lacking murine Ptpn22) exhibited a reduction in LPS-triggered type 1 IFN production, whereas inflammatory cytokine production remained intact. These studies potentially implicate Lyp620W in autoimmune pathogenesis by altering innate, as well as adaptive, immune responses.

Thus, studies of human cells, to this point, demonstrate an impact of the Lyp620W variant on both maturation and function of hematopoietic lineages, each of which may have the potential to contribute to autoimmunity. Notably, despite this information, it remains unclear which features, either singly or in combination, play a crucial role(s) in the events leading to disease. To answer these questions, murine models become increasingly important.

The function of Ptpn22 in murine lymphocytes was first extensively assessed in Ptpn22 knockout (KO) mice generated by the Chan group (25). Ptpn22 deficiency augments TCR signaling in CD4⁺CD8⁺ double-positive thymocytes and in peripheral effector/memory T cells, but not in the peripheral naive T cells, and these events result in enhanced thymic positive selection and expansion of peripheral effector/memory T cells, respectively, in Ptpn22 KO mice. Ptpn22 deficiency also leads to increased numbers of thymic and peripheral regulatory T cells (Tregs) (26, 27). Although Ptpn22-deficient Tregs exhibited normal suppressive activity in vitro (26), Brownlie et al. (27) found that they are more suppressive than WT Tregs in vivo, as evidenced by reduced expansion of effector T cells and increased suppression of colitis following adoptive transfer of Ptpn22 KO Tregs with WT effectors into Rag-deficient mice. Aged Ptpn22 KO mice exhibit increased numbers of spontaneous germinal centers (GCs) and increased serum Ab titers, events that are presumed to be regulated by B cell–extrinsic signals because BCR signaling and B cell development are intact in Ptpn22 KO mice (25). A recent study revealed that Ptpn22 deficiency increases GC activity, predominantly by increasing proliferation, survival, and cytokine secretion by T follicular helper (Tfh) cells (28). Ptpn22 KO mice on a nonautoimmune (C57BL/6 [B6]) background do not develop autoantibodies or overt autoimmunity (25). However, Ptpn22 deficiency, in cooperation with the CD45 E613R mutation, induces murine lupus-like autoimmune disease in B6 mice (16). Ptpn22 KO mice on the KBxN background have increased severity of arthritis (28), whereas Ptpn22 KO mice on the B6 background are resistant to experimental autoimmune encephalomyelitis (26), implying that the role of Ptpn22 in autoimmune pathology is dependent on the disease and genetic background of the murine model.

The impact of Ptpn22 in T1D pathology has been investigated using the NOD mouse model. Zheng and Kissler (29) generated transgenic NOD mice in which Ptpn22 is inducibly silenced by RNA interference. Notably, Ptpn22 knockdown (KD) mice exhibit reduced diabetic incidence compared with controls, indicating that Ptpn22 silencing has an overall protective effect against T1D. In related studies, Ptpn22 silencing resulted in hyperactivation and decreased survival in B cells. Conversely, Yeh et al. (30) reported that transgenic overexpression of WT Ptpn22 in T cells of NOD mice attenuated autoimmune diabetes in a dose-dependent manner. Ptpn22 overexpression led to a reduction in naive T cell TCR-mediated signals, inhibited autoreactive T cell proliferation in response to islet Ags, and led to decreased numbers of pathogenic pancreas-infiltrating T cells. One possible explanation for the discrepancy between these studies is a differential impact on the Treg compartment. Ptpn22 silencing leads to an expansion of peripheral Tregs, whereas Ptpn22 overexpression has no effect on Treg differentiation or suppressive activity. Alternatively, overexpression of Ptpn22 also may result in decreased activation of diabetogenic effector T cells.

To assess the mechanism by which the risk variant induces autoimmunity, three independent groups developed either knock-in (KI) mice expressing Ptpn22-R619W, a mutant analogous to the human variant, or transgenic mice overexpressing the human variant (13, 15, 31, 32). The Siminovitch group (31) generated Ptpn22-R619W KI mice on a B6 background. This strain exhibited phenotypic features similar to Ptpn22 KO mice, including lymphocyte and dendritic cell hyperresponsiveness consistent with a loss-of-function effect. Aged Ptpn22-R619W KI mice also developed enlarged spleens and thymuses, expanded effector/memory T cells, and spontaneous GCs. However, despite these features, Ptpn22 KI mice on the B6 background did not develop autoimmunity. Of note, the investigators also reported that both the murine and human variant proteins exhibited reduced stability due to enhanced calpain- and proteasome- mediated proteolysis, implying that loss of function occurs as a result of accelerated protein degradation.

Using a similar strategy, Ptpn22-R619W KI mice were generated and evaluated in a potentially more autoimmune-prone, B6 × 129 mixed genetic background (15). In contrast to the initial findings, the investigators showed that the murine and human variant proteins exhibited normal in vitro and in vivo stability, additionally supporting this conclusion using a Rosa26-transgenic model. Importantly, aged KI mice of the mixed genetic background developed autoantibodies and systemic autoimmunity, consistent with the predicted role for the risk variant in promoting disease. Further, using the Rosa26-transgenic model, the investigators demonstrated that B lineage–restricted variant expression was sufficient to mediate autoimmunity. Consistent with altered B and T cell homeostasis in human subjects carrying the risk variant (19, 24), KI mice developed expanded transitional B cell and effector/memory T cell populations (15). No significant alterations in Treg development or suppressive activity were observed in Ptpn22-R619W KI mice. Ptpn22-R619W enhanced TCR signaling in both naive and memory T cells and impacted a distinct subset of proximal TCR substrates. These latter data imply that, although the variant largely mimics a loss-of-function phenotype, the point mutation also uniquely alters Ptpn22 interactions, likely leading to variant-specific changes in function. Intriguingly, in contrast to naive or newly generated effector T cells, T cells isolated from aged KI mice exhibit attenuated TCR-induced calcium flux, indicating that the long-term outcome of enhanced signaling ultimately leads to hyporesponsiveness (X. Dai and D.J. Rawlings, unpublished observations). In line with this, as noted above, previous human studies revealed blunted AgR-mediated calcium flux in both memory T and B cells derived from subjects carrying the risk variant (19, 24).

The Peterson group (13) and the Bottini group (32) developed and reported two alternative transgenic models that overexpress human Lyp620W under the control of either the proximal Lck promoter or the human endogenous regulatory elements in BAC-transgenic mice, respectively. In the former model, thymocytes overexpressing Lyp620W exhibited reduced TCR signaling, findings that contrast with the augmented TCR signal observed in Ptpn22-R619W KI T cells. Also, in contrast to Ptpn22 KO or KI mice, thymic-positive and thymic-negative selection were not altered, suggesting that overexpression of the human PTPN22 risk variant was unlikely to alter murine T cell central tolerance. However, as noted with respect to the Jurkat T cell studies described above, potential nonphysiological effects due to altered protein:protein interaction following overexpression of the human variant must be considered when interpreting these phenotypes. Interestingly, myeloid cells isolated from human PTPN22 risk variant BAC-transgenic mice exhibit impaired TLR-related type 1 IFN production (13). Similar findings have not been reported using Ptpn22 KI mice, and the phenotype of T and B cells in the BAC-transgenic model has not been described.

A number of important observations linking human and mouse studies have emerged from recent work on the PTPN22 risk variant. One striking, albeit underappreciated, feature of human and murine studies has been the impact of the risk variant on B cell biology. Although early work focused on T cells, the near-complete association of the risk variant with diseases characterized by high-titer, disease-specific autoantibodies implies a crucial role for the variant in altering B cell function (19). Consistent with this idea, current data demonstrate B cell–intrinsic (as well as B cell–extrinsic) impacts on B lineage function. Indeed, one unifying feature of healthy individuals with the variant and the murine KI model is an alteration in the transitional B cell compartment (15, 24). These observations correlate with reduced in vivo apoptosis of early transitional B cells in the murine model and increased resistance to BCR-triggered apoptosis and enhanced CD40-mediated signals in vitro in human transitional B cells (23, 24). Although it remains unclear as to whether these observations reflect altered negative versus positive selection of developing B cells, additional studies using the murine KI model are anticipated to address this question. Another emerging theme is that, within an “at-risk” genetic setting, individuals with the variant or KI mice each can develop autoantibodies and autoimmunity over time, a concept that is best illustrated by the requirement for the mixed genetic background to promote disease development in the KI model. A combination of future studies, assessing cell signaling and phenotype in carefully genotyped healthy individuals and using cross-breeding of KI animals with strains bearing alterations in additional candidate genes, will be required to more fully understand events that synergize for loss of tolerance. Finally, both individuals with the variant and KI animals exhibit an expansion or skewing of effector memory T cells. As discussed in more detail below, we believe that this finding may reflect increased TCR signaling in response to self- or environmental Ags in both cases. However, it is interesting to note that T effector skewing is also observed in murine chimera models in which only the B cells express the variant protein (X. Dai and D.J. Rawlings, unpublished observations), implying that altered B cell tolerance can be sufficient to promote altered T cell activation.

What is different in human versus murine studies and why? The most perplexing observations when comparing the analysis of primary lymphocytes from human and murine studies have been the disparate response to AgR-triggered signaling. This difference is most evident in experiments using human memory T and B cells; individuals with the variant exhibit significantly blunted signals based upon both intracellular calcium flux and the induction of tyrosine-phosphorylated signaling effectors. These findings contrast with the exaggerated signals in thymocytes, naive T cells, or in vitro–expanded effector T cells derived from KI mice. Similarly, transitional and naive B cells isolated from humans with the variant exhibit blunted signaling that is not evident in primary B cells from young KI mice. Although these differences require definitive explanation, our observations in the KI model suggest that the reduced signaling in human memory and effector cells reflects the outcome of chronic Ag experience (present in both humans and mice) that occurs in the setting of an enhanced TCR signaling program in naive T cells. Our best evidence for this hypothesis is the blunted TCR signaling observed in T cells derived from aged KI mice (Fig. 1), cells that presumably experienced chronic Ag stimulation. Analogously, in the B cell compartment, we observed evidence for increased self-Ag–driven signaling in transitional B cells in the KI model. This conclusion is based upon evidence for increased positive selection of B cells expressing a phosphorylcholine-specific BCR in the KI model (15), as well as studies assessing the expression level of a BCR signaling–dependent reporter (16) (G. Mensler, X. Dai, and D.J. Rawlings, unpublished observations). Reporter expression is elevated in the KI model, implying that KI B cells experience increased self-Ag–driven BCR signaling during transitional development. We propose that this increased activation during development and Ag encounter also may help to explain the blunted signals observed in human immature and naive B cells. Admittedly, these observations remain preliminary, and alternative approaches are required to address this “signaling dichotomy.” Direct genome editing of PTPN22 in primary human lymphoid cells (leading to cell populations that differ only with respect to this single gene product) would provide a valuable means to address this key question. Alternatively, studies using truly naive human cells (such as those present in cord blood, in vitro T or B cell culture models, or humanized mice) may need to be used to accurately assess alterations in receptor signaling.

The observations reported in this article highlight the importance of seeking ways to optimally integrate investigations using human cell populations and murine models. Although work using human cells may identify breaches in tolerance [as demonstrated, for example, by elegant work showing the increased proportion of self-reactive naive B cells in carriers with the PTPN22 variant (23)], mechanistic questions regarding how the variant impacts development and repertoire selection remain challenging in humans. Linking KI models with other genetic or transgenic manipulations will be critical for such mechanistic insight; this approach has already begun to bear fruit, as evidenced by work showing the impact of the variant on thymic T cell selection and peripheral tolerance events in transitional B cells (15, 31). Similarly, human studies led to important insights via identification of alterations in cell phenotype, function, and signaling; such approaches are clearly amenable to an initial survey of combinatorial genetic phenotypes. As noted above, newly developed genome-editing tools have significant potential to improve studies in human cells. In addition, expanding technologies, including multiparameter flow cytometry or mass cytometry, cell type–specific RNA sequencing analysis, and mass spectrometry–based assessment of cell-specific signaling proteomes (33), will become increasingly important in these efforts. However, murine models and cross-breeding will still be required to further dissect such phenotypes, as illustrated by the “signaling dichotomy” described above. Another challenging question with respect to human work is dissection of the specific role for different cell types or developmental subsets in disease pathogenesis. Use of conditional expression of the Ptpn22 variant using the Rosa26 locus, murine cell transfers, and bone marrow chimera models have begun to address these questions. However, traditional murine KI strategies, including those used to generate the Ptpn22 KI mice, led to introduction of the variant allele in all lineages; therefore, they remain limited in their capacity to efficiently assess lineage-specific events in disease. Notably, several alternative genetic KI approaches have emerged, including the “FlEx” system using two pairs of heterotypic lox sites, to overcome these hurdles (34, 35). Thus, combining newer KI models with lineage-specific Cre-deletion systems will become increasingly valuable. Finally, an ongoing challenge of murine models will be the capacity to model noncoding variants and/or missense variants in which the SNP and coding residue may not be highly conserved, and, importantly, to do all of the above in a more timely fashion. Although not a complete solution, improved genome-editing tools and the ability to multiplex genetic manipulation in both murine and human cells will promote progress.

Another lesson highlighted in studies of the PTPN22 variant is that analyses using human cells must be considered in the context of how the risk variant may impact immune responses that occur following multiple forms of immune activation and over periods of many years. In vitro studies of human cord blood may yield insight into the functional phenotype seen in naive cells, whereas in vivo studies of responses to infection, vaccination, or biologic therapies may yield a better understanding of how a genomic variant may impact immune responses in a mature individual. Understanding how a risk variant can alter responses over time is vital because many autoimmune diseases develop in adults. Parallel murine studies will need to be designed to examine such outcomes. Our tools to address these questions using murine systems remain limited and would benefit from consideration of how a specific variant may have emerged within the human population including, in particular, its potential protective role(s) in immune responses to specific pathogens. Using this information may be helpful in both modeling relevant immunologic challenges and predicting lineages most likely to impact immune function in autoimmunity.

In summary, as detailed above and displayed in Table I, using a combined approach focused on human cells expressing the PTPN22 variant and parallel murine KI models, investigators in the field have contributed to an improved understanding of the role for the PTPN22 risk allele in immune tolerance, cell activation, and autoimmune pathogenesis.

Based on our current understanding of the impact of Lyp620W on the immune response, several mechanisms can be proposed to help explain its contribution to the development of autoimmunity (Fig. 1). Because the variant impacts multiple lineages and is linked to multiple distinct immune disorders, we propose that these processes likely function in concert to mediate a break in T and B cell tolerance and promote disease.

1. Disruption of normal tolerance checkpoints during peripheral B cell development leads to enrichment of a naive B cell repertoire with increased numbers of self- and polyreactive B cells, thereby increasing the potential to facilitate humoral autoimmunity.

2. Altered BCR signaling promotes self-Ag–driven activation of self-reactive B cells and leads to presentation of peptide self-Ags to autoreactive T cells; this process promotes the development of spontaneous GC reactions, resulting in generation of autoantibody-secreting plasma cells.

3. Altered TCR signaling modifies the TCR repertoire (effector and regulatory repertoire).

4. Possible increases in Tfh cell activity, leading to enhanced IL-21 production, may work in concert with altered B cell activation to promote or sustain autoimmune GC reactions.

5. Expansion of memory T cells and skewing of T cell differentiation may mediate direct tissue injury and facilitate dysregulated regulatory processes across a range of effector or memory populations.

6. Alterations in innate responses, including disrupted TLR-driven type I IFN production, might limit GC responses yet also promote a proinflammatory milieu that facilitates T cell activation and/or other events.

Despite this progress, a number of key unanswered questions remain. It will be important to identify the specific role(s) for the variant in B cell selection and shaping of the naive B cell repertoire. Further, it will be useful to determine whether specific self-reactive AgR TCR or BCR specificities are sufficient to trigger disease when expressed in primary cells carrying the variant allele. Such experiments would help to determine whether the variant’s impact on signaling can contribute to disease in a manner independent of its role in selection. The role of Lyp, in concert with Csk, in modulation of the activation state of Src family kinases is now well established. However, it is also likely that other protein:protein interactions are critical for Lyp function. Thus, it will be valuable to comprehensively identify these interactions and determine whether relevant signaling cascades are impacted by the variant. Finally, although the PTPN22 1858T variant has been reproducibly associated with multiple autoimmune diseases, it will be informative to perform more detailed sequencing analyses to determine whether rare PTPN22 variants exist within candidate disease cohorts and identify their impact in assays analogous to those studied to date.

Notably, one might argue that our mechanistic understanding of the variant has benefited from trying to reconcile disparate experimental findings obtained from alternative model systems. For example, discrepant findings regarding AgR signaling in cell lines (that overexpressed the variant protein with Csk versus without Csk) or in primary human versus murine lymphocytes has challenged investigators to be open to new ideas, including that altered signals may reflect a critical stoichiometry of protein interaction and that observed changes may reflect a previous history of receptor engagement and selection following immune activation, respectively. This work highlights how a single variant may impact multiple facets and functions of the immune system and, this, in fact, may be part of why it is a risk variant.

Finally, it is important to recognize that most genome-wide association study risk alleles, including PTPN22, exhibit relatively limited individual impact on disease development. Thus, a key argument for performing detailed analyses of a single risk allele (as performed for PTPN22) is that such work will identify shared “signatures” predicted to track with disease. These “signatures” may be driven by other genetic variants that cluster in common pathways. For example, multiple genes involved in BCR and TCR and other costimulatory signaling pathways have been implicated in autoimmune diseases that are also associated with the PTPN22 risk allele. Understanding the combined or synergistic impact of such variants represents a critical focus for future human and animal studies. We envision that these “signatures” will facilitate broader phenotypic studies designed to identify individuals at greater risk for disease and also highlight potential candidate therapeutic targets. Based upon this premise, we anticipate that murine modeling of additional common disease-linked variants (as well as of highly penetrant, but rare, dominant autoimmune mutations), paired with carefully designed human genotype:phenotype studies, will continue to be a highly productive avenue for mechanistic and therapeutic insight in human immune diseases.

The authors have no financial conflicts of interest.