The RAS/MAP kinase signaling pathway is central for cellular activation through numerous cell surface receptors. Furthermore, RAS is mutated in approximately 30% of all human tumors, making its study relevant to normal and cancer cell biology (1, 2). RAS can cycle between GTP- and GDP-bound forms and RAS guanine exchange factors (GEFs) facilitate conversion to the GTP-bound/active form of RAS. SOS1 is a prototypical RAS GEF. In T cells, activation through the T cell antigen receptor (TCR) results in RAS activation via SOS1 and other GEFs. RAS mediates downstream activation of the MAP kinases ERK1 and ERK2 (3, 4). Four RAS GEFs are active in developing and mature T cells: SOS1, SOS2, RASGRP1, and RASGRP4. Their relative expression levels determine their contributions to thymocyte development and T cell activation (3).

Genetic analysis has been essential for investigating the roles of SOS proteins in T cells. Investigation of SOS1 function was impeded when SOS1 deficiency in mice was found to be embryonic lethal (5). Kortum et al. generated floxed (conditional) Sos1-deficient mice to address the function of SOS1 in specific cell lineages (6). By crossing Sos1^f/f mice to Lck-Cre transgenic mice, SOS1 deficiency was achieved in developing and mature T cells. T cell development was partially disrupted. The effect on T cell development was mediated by both scaffolding and GEF functions of SOS1 (7). To bypass the effects of SOS1 on T cell development and to study the effects of SOS1 deficiency on mature T cells, we crossed Sos1^f/f mice with CD4-Cre transgenic mice (8). In addition, those mice were bred to mice deficient for the related Sos family member SOS2 (9). Sos1^f/f Sos2^−/− CD4-Cre⁺ mice had normal numbers and subsets of T cells. T cells from Sos1^f/f Sos2^−/− CD4-Cre⁺ mice had intact ERK activation downstream from the TCR but defective ERK activation downstream from the IL-2 receptor. The Sos1^f/f Sos2^−/− CD4-Cre⁺ mice displayed defects in T cell migration that were secondary to increased PI3kinase activity in T cells (8).

For the latter studies, CD4-Cre was chosen for conditional knockout of Sos1 because it is expressed later in thymocyte development than Lck-Cre (10). Lck-Cre is expressed at the early CD4 CD8 double negative stage whereas CD4-Cre is expressed at the transitional CD4 CD8 double positive stage. CD4-Cre is widely used for conditional deletion of genes in double positive and mature T cells as the original description of CD4-Cre (11) has been cited over 500 times. Recently the integration of the CD4-Cre transgene, which expresses the Cre recombinase gene under the control of the promoter and regulatory elements of the Cd4 gene, has been mapped to mouse chromosome 3 and is present in at least 15 copies (12).

In addition to the signaling effects described above that we observed in mature T cells from Sos1^f/f Sos2^−/− CD4-Cre⁺ mice, we also observed that the mice developed nodules on multiple joints as they aged, especially carpal joints, and could eventually develop hind limb paralysis and become lame. We postulate that these disabling defects arise from abnormal cartilage homeostasis. Our unexpected results of a cartilage-based phenotype in Sos1^f/f Sos2^−/− CD4-Cre⁺ mice provide evidence that CD4-Cre expression may not be restricted to conventional double positive and mature CD4⁺ and CD8⁺ T cells.

To investigate the effects of Sos1 and Sos2 deletion in T cells, we crossed Sos1^f/f Sos2^−/− mice to CD4-Cre transgenic mice. Expression of Cre recombinase in CD4 CD8 double positive thymocytes, in single positive thymocytes, and in mature CD4⁺ and CD8⁺ T cells results in an absence of SOS protein expression in these cells, which is consequent to the deletion of Sos1 exon 10 by Cre recombinase (6). We were surprised, then, to observe that Sos1^f/f Sos2^−/− CD4-Cre⁺ mice developed visible nodules on multiple joints, especially carpal joints (Figure 1A, arrows) and had reduced joint flexibility by about 16 weeks of age. 3D CT imaging of a 32-week-old Sos1^f/f Sos2^−/− CD4-Cre⁺ mouse showed bony protrusions on the carpal joints and severe kyphosis of the spine (Figures 1B–D; Videos S1 and S2 in Supplementary Material). As the mice aged, some exhibited hind limb paralysis and became lame. Figure 1E shows clinical scoring based on nodule presence on joints, joint flexibility, and gait. An observer blinded to the genotypes scored the mice at 20 and 26 weeks of age. The observer examined the following four genotypes: WT (Sos1^f/f Sos2^+/+ CD4Cre⁻), Sos1 KO (Sos1^f/f Sos2^+/+ CD4Cre⁺), Sos2 KO (Sos1^f/f Sos2^−/− CD4Cre⁻), and Sos1/2 dKO (Sos1^f/f Sos2^−/− CD4Cre⁺). With the exception of one Sos2 KO mouse, the only mice that scored above a normal clinical score of 1 were Sos1/2 dKO mice. At 26 weeks of age, 16 of 17 Sos1/2 dKO mice had measurable clinical manifestations of disease and 12 of 17 Sos1/2 dKO mice had severely reduced joint flexibility and were lame (clinical score greater than or equal to 4).

To further characterize the abnormal structures that we observed clinically, we performed histological analysis on the vertebrae, carpal joints, and stifle joints of Sos1/2 dKO and control Sos2 KO mice. Figure 2 shows Hematoxylin and Eosin staining and Toluidine blue staining of vertebrae and carpal joints. Toluidine blue staining, which preferentially stains cartilage blue, showed increased amounts of abnormally organized cartilage in these joints in Sos1/2 dKO mice, but not Sos2 KO mice. By 12 weeks of age, the growth plates of Sos1/2 dKO mice showed proliferation of dysplastic chondrocytes and these changes became more prominent over time. By 25 and 39 weeks, dysplastic chondrocyte proliferation had resulted in irregular and sometimes severe thickening of the physes that multifocally impinged on adjacent structures, including the spinal cord. Spinal cord compression likely contributed to the hind limb paresis and altered gait seen clinically in these mice. Importantly, there was no evidence of lymphocyte infiltration or inflammation in Sos1/2 dKO sections.

To better understand the tissue architecture of the carpal nodules, we performed SHG imaging of Sos1/2 dKO carpal nodules and the corresponding area of control Sos2 KO carpal joints. SHG enables the label-free visualization of collagen fibers in various tissues, as previously shown (13). A loosely organized pattern of wavy collage fibrils was observed in control carpal joints (Figures 3A,C). In contrast, the pattern in the Sos1/2 dKO carpal nodules contained relatively straight collagen fibrils that were closely aligned (Figures 3B,D). SHG imaging of edges of the nodules indicated that the nodules seemed to be encased in a rich collagen network (Figure 3E).

Because T cell infiltration was not seen histologically in carpal nodule sections (Figure 2C), we postulated that some other non-T cells in the Sos1/2 dKO nodules might have deleted Sos1 exon 10. To investigate Sos1 deletion in carpal nodules, we homogenized nodule tissue and performed a genomic PCR that distinguishes between wild-type Sos1, floxed Sos1, and deleted Sos1 (Figure 4). Lanes 5 and 7 show that nodule tissue does indeed contain cells with Sos1 exon 10 deletion. We also spiked normal tissue with known numbers of mature, deleted CD4⁺ T cells derived from the lymphoid organs of Sos1/2 dKO mice. These samples showed that greater than 10⁴ T cells/mg tissue would be required to see a deleted signal such as that seen in lanes 5 and 7. Examination of the pathology of Sos1/2 dKO carpal tissue indicates that it is extremely unlikely that the deleted PCR band originating from carpal tissue arises exclusively from T cells because lymphocyte infiltration was not observed in nodule sections.

To investigate whether normal chondrocytes express Cd4 and, therefore, might be susceptible to CD4-Cre-mediated deletion, we performed quantitative real-time RT-PCR on growth plate cartilage from 1-week-old wild-type mice (Table 1). Levels of Cd4 expression were extremely low compared to the positive controls from lymph node single cell suspensions and purified CD4⁺ lymph node T cells. However, the levels of Cd4 expression in growth plate cartilage were slightly higher than Cd3e levels.

To more definitively determine if mature, conventional CD4⁺ and CD8⁺ T cells, which are the usual intended target of CD4-Cre transgenesis, were responsible for cartilage disease and nodule formation in Sos1/2 dKO mice, we crossed Sos1/2 dKO mice to mice deficient in RAG2, a recombinase that is required for the formation of mature T and B lymphocytes. As shown in Table 2, all Sos1/2 dKO Rag2^−/− mice developed carpal nodules at 18–20 weeks similar to Sos1/2 dKO Rag2^+/+ mice. At 26 weeks, 6 of 6 Sos1/2 dKO Rag2^−/− mice had a maximum clinical score of 6 whereas 2 of 4 Sos1/2 dKO Rag2^+/+ mice had a clinical score of 6. These data indicate that mature, conventional TCR-expressing T cells are not required for development of the cartilage-based disease described in this study. In fact, in the absence of conventional T and B lymphocytes, clinical disease was more severe.

To investigate the role of SOS proteins in T cells, we crossed Sos1^f/f Sos2^−/− mice with CD4-Cre transgenic mice. We previously reported no effect on T cell numbers, but an effect of these mutations on T cell signaling and T cell migration (8). Unexpectedly, as the dKO mice aged, we observed nodules in the joints of greater than 90% of these mutant mice, especially at carpal joints. Histological analysis suggested that the abnormal growth originated from dysplastic chondrocytes. An important question is whether the joint abnormality that we observed is caused by T cells, as would be expected from using CD4-Cre, or by other non-T cells expressing CD4-Cre sometime in their lifetime. Histological analysis did not show lymphocytic infiltrates in the nodules. Furthermore, crosses to Rag2^−/− mice indicated that the phenotype was not caused by T cells, at least not conventional T cells that undergo rearrangement of the TCR, because nodules still formed in the joints of the Sos1/2 dKO Rag2^−/− mice.

What is the origin of the chondrocyte dysplasia in Sos1/2 dKO mice? One possible scenario is that the cartilage nodules arise from impaired SOS/RAS/MAPK signaling in a rare subset of chondrocytes consequent to CD4-Cre expression in those chondrocytes. Analysis of Cd4 RNA expression in normal chondrocytes from growth plate cartilage showed low, but detectable, expression leaving open the possibility that CD4-Cre is active in a small subset of chondrocytes. This scenario is consistent with previous studies demonstrating that the RAS signaling pathway, mediated by SOS, inhibits chondrocyte proliferation (14). In chondrocytes, a fibroblast growth factor, FGFR3, signals through SOS to activate the RAS/MAPK pathway. Downstream from RAS, RAF activates MEK1/2 and MEK5/6, which activate ERK and p38, respectively. Postnatal chondrocyte-specific Fgfr3 deletion results in chondroma-like lesions (15). Other RAS/MAPK perturbations that result in bone growth abnormalities include constitutively active MEK1 (16) and conditional inactivation of Erk1 and Erk2 in chondrocytes (17). Similarly, conditional knockout of Shp2 in chondrocytes results in the formation of cartilaginous nodules, similar to the phenotype seen in the current study. SHP2 is a protein tyrosine phosphatase that mediates RAS/MAPK signaling downstream from many growth factor receptors including FGFR (18, 19).

In humans, patients with Noonan syndrome have inherited mutations in RAS/MAPK pathway genes. Noonan syndrome is a multisystem and varied disease often characterized by distinctive facial features, developmental delay, short stature, congenital heart disease, and other abnormalities (20). About 10% of Noonan syndrome patients have SOS1 mutations that are thought to confer hyperactivity of SOS1. The short stature in these patients suggests that activating mutations in SOS1 impair growth plate chondrogenesis, whereas in our mouse model, loss of SOS1 and SOS2 appears to increase chondrogenesis.

Although it is possible that some rare chondrocytes express Cd4 and that chondrocyte dysplasia results from SOS deficiency in those chondrocytes, it is also possible that conventional T cells regulate chondrocyte proliferation and that SOS deficiency in T cells results in alterations in secretion of factors that normally keep chondrocyte proliferation in check. This cannot be the only mechanism at play, however, because Rag2^−/− mice (with wild-type SOS) do not have conventional T cells and do not exhibit a cartilage phenotype. Another possible scenario is that immune regulatory cells, which express Cd4 sometime in their lifetime, but are not conventional T or B lymphocytes, keep chondrocyte proliferation in check. Consequent to SOS deficiency in these cells, chondrocyte dysplasia would result, disturbing cartilage homeostasis. Our results that 26-week-old Sos1/2 dKO Rag2^−/− mice have more severe disease than Sos1/2 dKO Rag2^+/+ mice could indicate that imbalances in immune cell subpopulations can alter cartilage homeostasis. At the least, these results imply that the disease is not completely due to a chondrocyte effect. Some combination of these possible causes of chondrocyte dysplasia, SOS deficiency in rare chondrocyte populations, in T cells, or in regulatory non-T cells, may be responsible for development of this clinical disease. Control of cartilage homeostasis by T cells and/or by hypothetical immune, non-T cells would reveal a novel function for the immune system.

The extensive prior evidence that SOS/RAS/MAPK pathway signaling inhibits chondrogenesis provides for the possibility that Sos1 is excised by CD4-Cre in occasional chondrocytes or chondrocyte precursors. In combination with the germline loss of Sos2, RAS/MAPK signaling could be impaired, stimulating chondrocyte proliferation and nodule formation. Alternatively, Sos1 could be deleted in other cell types that regulate chondrogenesis via the secretion of cytokines or growth factors. In accordance with the hypothesis that CD4-Cre-regulated deletion of SOS/RAS/MAPK pathway genes can lead to the cartilage-based phenotype described here, the laboratory of Maureen McGargill has found that mice with both a germline mutation of Erk1 and conditional deletion of Erk2 mediated by CD4-Cre exhibit a similar disease to that described in this study (Marie Wehenkel and Maureen McGargill, St. Jude Children’s Research Hospital, Memphis, TN, USA, personal communication). As ERK kinases are MAPKs activated downstream from SOS proteins, it is extremely likely that the SOS/RAS/MAPK pathway causes the phenotype described here. Our results serve as a cautionary tale to those intending to use CD4-Cre transgenic mice. In addition to being expressed in conventional T cells, CD4-Cre may be expressed in a subset of non-T cells, either immune lineage cells (e.g., non-conventional T cells) or non-immune lineage cells (e.g., chondrocytes). This expression may give rise to the dramatic phenotype observed here, depending on the gene being deleted by the Cre recombinase. Disruptions of genes in the RAS signaling pathway are particularly likely to render this phenotype.

GG, JB, RW, JK, LS, and CS designed the study. GG, DG, WL, NM, JL, RK, NS, SH, and CS performed the research. JB, RW, JK, LS, and CS supervised the study. CS, GG, JB, RW, JK, and LS wrote the paper.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.