SHP-2 KO mice develop tumors with a marked wrist location tropism
Using mice conditionally deficient for SHP-2 in the T cell lineage, we previously reported that the development and function of these lymphocytes is globally intact 13. However, we also reported that in aging mice, Ptpn11 gene deletion driven by CD4 Cre recombinase leads to wrist tumors in a T cell-independent manner 13. To mark cells that have deleted the Ptpn11 gene, we crossed the SHP-2fl/fl-CD4Cre mice to RosaYFP reporter mice 32. For simplicity, we will refer to SHP-2fl/fl-EYFP+-CD4Cre animals as KO, and their littermate controls SHP-2fl/--EYFP+-CD4Cre as Het, throughout. We previously reported that tumors were observed in the wrists (>3 months of age), and sporadically in the hips, knees, and vertebrae of older mice (>5 months of age) 13. Upon further examination, the tumors were mostly identified in bones formed by endochondral ossification, specifically long bones and vertebrae. We next monitored wrist tumor development at 1, 3, and 6 months of age. Tumors were absent in Het animals at each timepoint examined and mostly undetectable in 1-month-old KO animals (Supplemental Figures 1A). At 3 months of age, wrist tumors were identified in all KO animals, and tumors continued to progress in size at 6 months of age, becoming visible in gross images (Supplemental Figure 1A). When measured, wrists were significantly wider in 6-month-old KO animals as compared to Het animals (5.5 ± 0.5 mm versus 1.5 ± 0.4 mm; p<0.05, Supplemental Figure 1B). Tumor growth was multidimensional and observed in the distal head of the ulna and radius but not in the proximal ends (Supplemental Figures 1A). Interestingly, tumors were not observed in the wrist carpels or meta carpels and did not develop along the shaft of the long bones. However, some of the animals developed non-wrist tumors in other joints (but not all), that were uni- and bilateral (Supplementary Figure 1C). Additional tumors could be seen in the hip (proximal femur), and knee (distal femur and proximal tibia), but were never observed in the ankle (distal tibia and fibula) and feet (Supplementary Figure 1C). In contrast to the wrist tumors, tumors in other joints were mostly small. Altogether, these data show that SHP-2 deletion in CD45-CD4+ cells (YFP+) lead to tumor development with a marked wrist location tropism.
Non-hematopoietic CD45-YFP+ cell number increased during wrist tumor development
We previously demonstrated that Ptpn11 gene deletion driven by CD4 Cre recombinase leads to cartilage tumors in a T cell independent manner 13. However, the cell subset responsible for this phenotype was not characterized. To investigate this, we first examined cells from the contralateral wrists of 1-, 3- and 6-month-old Het and KO animals. Surprisingly, we identified a non-hematopoietic CD45-YFP+ cell subset which progressively increased in frequency and number in the wrists of KO animals as they aged (Figure 1). These increases were not present in Het animals and became significant between the two groups at 6 months of age (p < 0.05). Significant increases, of a lesser magnitude, in percentage and number of CD45-YFP+ cells were also identified in the pooled bone ends of 6-month-old KO animals as compared to Het animals (Supplemental Figures 2A & B). These changes were not identified in the central bone marrow or spleen, where only very low frequencies (<0.05%) were identified at any timepoint in Het or KO animals (Supplemental Figures 2C & D).
CD45-YFP+ cells isolated from tumors express little CD4 and have a mesenchymal phenotype
We performed an extensive characterization of the CD45-YFP+ cells associated with the development of wrist tumors. At 6 months of age, these cells expressed limited amounts of surface CD4, no intracellular CD4, and had high levels of YFP expression by flow cytometry (Figure 2A). These findings were confirmed by examining the mean fluorescent intensity (MFI) of CD45-YFP+ cells compared to CD45-YFP- and CD45+YFP+cells isolated from tumors (Figure 2B). For positive and negative MFI controls, CD45+YFP+ cells (largely T cells) were further subdivided by positive surface CD4 (CD4+) and negative surface CD4 (CD8+) expression. Based on MFI, CD45-YFP+ cells had only modest CD4 surface expression (<500), compared to CD45+YFP+CD4+ cells (>6500), and CD45+YFP+CD4- (<25). Intracellular CD4 was <50 for all subsets examined except for CD45+YFP+CD4+ cells which were >700. The YFP expression was found to be highest in CD45-YFP+ cells (~4500), compared to both CD45+YFP+ subsets (~1500) (Figure 2B). Interestingly, the CD45-YFP+ tumor-associated cells were found to be negative for all hematopoietic markers examined (Figure 2C). This included a lineage cocktail (Ter119 (erythrocytes), LY6C (granulocytes), CD11b (macrophages), and B220 (B-cells)), CD34, and C-Kit, while few expressed Sca-1. Notably, these cells expressed the mesenchymal markers integrin b1 (CD29) as well as integrin b6 (CD49f), and were found to have significantly greater forward scatter and side scatter compared to CD45-YFP- and CD45+YFP+ cells (Figure 2C). Taken together, these data indicate that the CD45-YFP- cells are derived from mesenchymal origin.
CD45-YFP+ cells have an arrested stem cell phenotype
Having determined that the tumor-associated CD45-YFP- cells were from mesenchymal origin, we then continued the characterization of these cells. Phenotypic expression of SSC markers has been used to identify chondrocyte precursor cells 19. Wrist cells from adult (3-5 months of age) Het and KO animals were extensively phenotyped to more accurately define the cells associated with tumors (Figure 3). First, lineage negative cells (CD45-, TER119-, 6C3-, CD202b-) were divided into YFP+ and YFP- subsets. While approximately 2/3 of the lineage negative cells isolated were YFP- in Het mice (Figure 3A), the inverse was found in KO animals with approximately 4/5 of the cells being YFP+ (Figure 3B). Lin-YFP+ and Lin-YFP- subsets isolated from Het and KO animals were mostly CD51+CD90- (73-95%), suggesting an SSC origin and disruption of developmental progression. Both YFP+ and YFP- cells were examined for the expression of markers according to the following progressive development schema; SSC (CD200+CD105-), pre-BCSP Cell C (CD105-CD200-), BCSP (CD105+CD200-), and PCP (CD105+CD200+). When Lin-YFP+ cells from tumors were further characterized using anti-CD105 and anti-CD200 mAbs, we found that this subset was largely composed of a pre-BCSP (CD105-CD200-) or BCSP (CD105+CD200-) phenotype with few cells expressing a SSC (CD200+CD105-) and almost none expressing a PCP (CD105+CD200+) phenotype (Figure 3C & D, left two columns). Within the Lin-YFP- cell population, larger percentages of SSC and PCP cells were identified (Figure 3C & D, right two columns). Importantly, YFP+ cells never progressed to the PCP phenotype and few CD90+YFP+ cells were identified. Altogether, these data demonstrate that CD45-YFP+ cells are mostly negative for lineage markers and express markers associated with mesenchymal stem cells and chondrocyte subsets. We concluded that YFP+ tumor associated cells have a BCSP cell phenotype.
YFP+ cells are observed in the growth plate of both Het and KO young animals
Immunofluorescent images of 1-month-old animals revealed the presence of YFP+ in the growth plates of the distal ulna and radius in the KO animals as expected but also in the heterozygote animals (Figure 4). YFP+ cells were identified in both the resting and proliferative zones (PZ) of the growth plate (GP) which were marked by the chondrocyte columns. The chondro-osseous junction was delineated with an anti-CD44 mAb, which stains osteocytes and hypertrophic chondrocytes 33 located beneath the chondrocyte columns. The YFP+ cells were only sporadically observed in the hypertrophic zone/ossification center (OC). In agreement with the flow cytometry, the YFP+ cells found in the growth plate did not co-stain with CD45 or TCRb, whereas the YFP+ cells found in the ossification center were largely CD45+ and TCRb+ (data not shown). Overall, the presence and location of these cells in the growth plates of KO and Het animals were found to be comparable at 1 month of age (Figures 4A-D). YFP+ cells were also identified in the growth plates of other bones examined, with less frequency but comparable location. In 1-month-old KO animals, YFP+ cells were found in the chondrocyte columns of the growth plate within the distal femur but were more diffuse than the distal ulna or radius (Supplemental Figure 3). Few to no YFP+ cells were identified in the proximal ulna or radius (elbow) of 1-month-old animals examined (data not shown). These data demonstrate that the YFP+ cells in adolescent Het and KO animals are found in comparable locations in normally structured growth plates.
YFP+ cells are still present in growth plate-like structures of 6-month-old SHP-2 KO animals
YFP+ cells were identified in the remnants of the growth plate and still aligned in a columnar fashion in 6-month-old Het animals (Figures 4E & F). It was noted that the columns contained far fewer cells as compared to 1-month-old animals. In addition, ossification centers marked by CD44 were not observed in 6-month-old Het animals. These animals had disorganized end bone architecture within the tumor. There were still active growth plate-like structures identified throughout the tumors of KO animals, which were marked by YFP+ cells in chondrocyte columns that appeared to be multidirectional (Figures 4E & F). Importantly, an erratic chondro-osseous junction was noted at the base of the columns where CD44+ cells were still present in all 6-month-old KO tumors examined. Notably, large numbers of YFP+ cells were associated with the multi-directional growth plates but largely absent from ossification centers (Figures 4G & H). To examine the presence of T cells in the tumors, sections were stained with an anti-TCRb mAb. TCRb+YFP+ cells were not observed in the growth plate-like areas, but clusters of T cells were seen beneath the ossification centers (Supplemental Figure 3).
SHP-2 regulates SOX-9 function in CD4+ cells from chondrocyte origin
SOX-9 is required in several successive steps of the chondrocyte differentiation pathway during endochondral bone formation in vivo, to ensure chondrocyte differentiation and prevent development towards osteoblastic differentiation (Figure 5A). Therefore, SOX-9 is absolutely necessary for chondrocyte specification and early differentiation 34. We hypothesized that SHP-2-deficient cells may either differentiate into chondrocyte-like cells or cause neighboring SHP-2-sufficient chondrocytes to contribute to exostoses and enchondromas via paracrine signals. To test this hypothesis, we generated SOX-9fl/fl-SHP-2fl/flCD4Cre (DKO) mice and SOX-9fl/+-SHP-2fl/flCD4Cre control mice. In the SOX-9 heterozygote controls (SOX-9fl/+-SHP-2fl/fl-CD4Cre), one copy of SOX9 is expressed, but Ptpn11 is deleted from CD4+ cells. As shown in Figures 5B & C, 100% of these mice developed tumors, indicating that expression of one copy of the Sox9 gene is sufficient for the phenotype. In contrast, none of the double deficient animals (Sox9fl/fl-SHP-2fl/fl-CD4Cre) developed tumors (Figures 5B & C). This demonstrated that SHP-2 intrinsically regulates SOX-9 expression in cells with a history of CD4 expression, and that SOX-9 expression is “unleashed” in the absence of SHP-2 (see model, Figure 5E).