Altered endochondral ossification in collagen X mouse models leads to impaired immune responses

Abstract

Disruption of collagen X function in hypertrophic cartilage undergoing endochondral ossification was previously linked to altered hematopoiesis in collagen X transgenic (Tg) and null (KO) mice (Jacenko et al., 2002). Mice displayed altered growth plates, diminished trabecular bone, and marrow hypoplasia with an aberrant lymphocyte profile throughout life. This study identifies altered B220⁺, CD4⁺, and CD8⁺ lymphocyte numbers, as well as CD4⁺/fox3P⁺ T regulatory cells in the collagen X mice. Additionally, diminished in vitro splenocyte responses to mitogens and an inability of mice to survive a challenge with Toxoplasma gondii, confirm impaired immune responses. In concert, ELISA and protein arrays identify aberrant levels of inflammatory, chemo-attractant, and matrix binding cytokines in collagen X mouse sera. These data link the disruption of collagen X function in the chondro-osseous junction to an altered hematopoietic stem cell niche in the marrow, resulting in impaired immune function.

INTRODUCTION

Hematopoietic disorders such as bone marrow failure, immune dysfunction, as well as immuno-osseous disorders may involve a direct link between the endochondral skeleton and the marrow. Thus, understanding how a marrow environment is established, what makes this site preferential for hematopoiesis, and which skeletal defects may contribute to marrow alterations and impaired hematopoietic differentiation, may be key for the diagnosis and treatment of various immuno-osseous, as well as hematopoietic diseases.

During vertebrate development, hematopoiesis becomes sequentially re-established in several anatomically distinct sites. However, the bone marrow represents the predominant site of blood cell production after birth (Aguila and Rowe, 2005) and is developmentally linked to endochondral ossification (EO) (Chan and Jacenko, 1998; Lefebvre and Smits, 2005). During EO, the generation of a cartilage anlage serves as a blue print for the axial and appendicular skeleton, as well as for certain cranial bones, and identifies all future skeletal elements where a functional marrow space could form. The gradual replacement of the cartilage anlagen by trabecular bone and marrow first involves a well-orchestrated series of chondrocyte maturation events, culminating in chondrocyte hypertrophy. This latter process results in a dramatic increase of cell size, concomitant with synthesis of a unique extracellular matrix (ECM) consisting predominantly of collagen X. Elaboration of the hypertrophic cartilage matrix, along with a repertoire of growth and signaling factors, permits vascular invasion and influx of mesenchymal cells, hematopoietic precursors, and osteo/chondroclasts. Through the action of these cells, the hypertrophic cartilage becomes degraded, while remnants of the cartilage matrix serve as scaffolds upon which osteoblasts deposit osteoid, thus forming trabecular bone at the chondro-osseous junction. As these events occur, a marrow cavity is formed in the primary followed by the secondary ossification centers, which define the cartilaginous growth plate of the developing bone.

Based on studies involving collagen X mouse models where collagen X function was disrupted in the growth plate either by transgenesis leading to dominant interference (Tg mice; (Jacenko et al., 1993a; Jacenko et al., 2001; Jacenko et al., 2002), or through gene inactivation by targeting and homologous recombination (KO mice; (Rosati et al., 1994; Gress and Jacenko, 2000), we have long maintained that skeletal development and hematopoiesis are intimately linked. All mice with altered collagen X function developed a variable skeleto-hematopoietic disease phenotype, ranging from perinatal lethality at week-3 (~25% in Tg and ~10% in KO mice; in these studies “perinatal” denotes days within the first month after birth), to either transient dwarfism or no discernable outward disease phenotype. Aging surviving mice, however, were susceptible to skeletal deformities, skin ulcerations, and aggressive lymphosarcomas (Jacenko et al., 2002). Skeletal defects in all collagen X mice involved EO-derived tissues, and included growth plate changes (e.g. compressed growth plates with a diminished hypertrophic zone in Tg, and thinned growth plates with a reduced proliferative zone in KO), decreased number and length of trabecular bony spicules, and generalized osteopenia (Kwan et al., 1997; Gress and Jacenko, 2000; Jacenko et al., 2002). Hematopoietic defects included marrow hypoplasia, altered lymphocyte development in marrow, and aberrant B and T cell levels in lymphatic organs (Gress and Jacenko, 2000; Jacenko et al., 2002). In general, a more severe skeleto-hematopoietic phenotype has been described in the collagen X Tg mice as opposed to the KO mice; this is consistent with the dominant interference mechanism of the transgene as opposed to protein depletion (Chan and Jacenko, 1998; Jacenko, 1998; Gress and Jacenko, 2000; Jacenko et al., 2001). Likewise, in patients with Schmid Metaphyseal Chondrodysplasia (SMCD), caused by mutations in the carboxyl terminal domain of collagen X, the skeletal disease manifestations were also more severe than those reported in the collagen X KO mice, owing to the gain-of-function effect of the mutations causing an ER-stress response in the chondrocytes (Ho et al., 2007). In both the collagen X Tg and KO mice however, the most acute disease manifestations were always observed in the perinatal lethal mice that exhibited hunching of back, lethargy, dehydration, wasting, and death ensuing within hours of the visible onset of the disease phenotype. In the perinatal lethal subset, growth plates were compressed, trabecular bone was sparse, marrows were aplastic with pronounced lymphopenia, and lymphatic organs were atrophic ((Gress and Jacenko, 2000; Jacenko et al., 2002); Sweeney et al., manuscript in preparation).

These disease manifestations of the collagen X Tg and KO mice are a direct result of disrupted collagen X function, since transgene insertional mutagenesis (Jacenko et al., 1993a; Jacenko et al., 1993b), mis-expression of endogenous or mutated collagen X in extra-skeletal tissues (Campbell et al., 2004), as well as contributions of strain-specific gene modifiers of either the C57BL/6 or DBA/2J mouse strains (Sweeney et al., manuscript in preparation) have been excluded. This study extends our analyses of altered hematopoiesis in the collagen X mice by addressing lymphocyte dysfunction via B and T lymphocyte counts and functional assays involving in vitro and in vivo challenges of the immune response. Data presented here support impaired immune function and aberrant levels of several inflammatory, chemo-attractant, and matrix binding cytokines in the collagen X mice as a consequence of disrupted collagen X function at the chondro-osseous junction in EO-derived bones. Taken together, these data suggests a provocative hypothesis linking EO to the establishment of a properly functioning hematopoietic stem cell (HSC) niche.

RESULTS

Altered collagen X function effects lymphocyte numbers

Previous studies have shown that size, architecture, as well as lymphocyte percentage of spleens and thymuses from collagen X Tg and KO mice were abnormal when compared to wild type cohorts, with the most extreme differences observed in the perinatal lethal subsets (Gress and Jacenko, 2000; Jacenko et al., 2002). To assess lymphocyte availability in the collagen X mice, the total number of thymocytes and splenic lymphocytes were determined by manual count and flow cytometry. When compared to controls, similar numbers of thymocytes, but increased numbers of B220⁺, CD4⁺ and CD8⁺ splenocytes (B and T cell markers respectively) were enumerated in the collagen X KO mice (Fig. 1), which exhibit ~10% perinatal lethality at week-3 (Gress and Jacenko, 2000). Conversely, the collagen X Tg mice, which exhibit ~25% perinatal lethality (Jacenko et al., 2002), displayed reduced numbers of both thymocytes and splenic lymphocytes when compared to wild type controls (Fig. 1) Further, the perinatal lethal subset of both the collagen X Tg and KO mice presented with drastically diminished lymphocyte populations in the thymus and spleen when compared to wild type mice (Fig. 1). Of note, the trends in lymphocyte number did not change once data were normalized to the spleen and thymus size (data not shown). Lastly, the collagen X Tg mice, as well as the perinatal lethal subset of Tg and KO mice, had significantly reduced numbers of T regulatory cells (Tr) (Fig. 2) which are implicated in regulation of inflammatory and autoimmune T cell activity (Mottet and Golshayan, 2007).

Figure 1. Collagen X mice have altered lymphocyte numbers.

Cell count of thymocytes (A), B220⁺ B splenocytes (B), CD4⁺ T splenocytes (C) and CD8⁺ T splenocytes (D) from wild type (WT), collagen X null (KO), collagen X transgenic (TG), and Tg and KO perinatal lethal (mut) mice. Number of thymocytes was counted directly, and number of B and T splenocytes was derived from the total number of live cells multiplied by the antibody specific percentages via flow cytometry. Note a decrease in lymphocyte number in TG, Tg mut, and KO mut mice, but an increase in the KO (n= number of mice per group, *P<0.05, **P<0.001).

Figure 2. CD4⁺/fox3P⁺ T regulatory cells are reduced in collagen X transgenic and perinatal lethal mice.

Splenocytes from week-3 wild type (WT), collagen X null (Col X KO), collagen X transgenic (Col X Tg) and collagen X Tg perinatal lethal mice (Col X mut) were stained with antibodies toward CD4 and fox3P to quantify T regulatory cells. (7 mice per group, *P<0.05, **P< 0.001).

Lymphocytes from collagen X mice have impaired responses to non-specific mitogens

The above data implied that altered collagen X function at the chondro-osseous junction results in the reduction of available lymphocytes to mount an immune response (Fig. 1 & Fig. 2). Thus to assess the quality of the lymphocytes present in the collagen X mice, in vitro lymphocyte stimulation assays were performed. For this purpose, non-specific mitogens concanavalin A (Con A) and phorbol 12-myristate 13-acetate/Ionomycin (PMA/I) were used to stimulate lymphocytes via the T cell receptor (TCR)/CD3 complex (Con A), or intracellularly via activation of protein kinase C (PKC) and Ca⁺⁺ mobilization (PMA/I) from collagen X and wild type mice at week-3 and over five months of age. Adult collagen X mouse splenocytes were stimulated to similar levels as wild type mice in response to PMA/I, as measured by secretion of interferon-gamma (IFN-γ) (Fig. 3D). However, adult collagen X mouse splenocytes stimulated with Con A responded with half of the population secreting less IFN-γ and half exhibiting background levels of IFN-γ compared to wild type controls, consistent with a defect in Con A responsiveness (Fig. 3B). Similar to adult mice, week-3 collagen X Tg and KO mice had normal secretion of IFN-γ in responses to PMA/I, but significantly reduced response to Con A (Fig. 3 A, C). In contrast, collagen X Tg and KO perinatal lethal mouse splenocytes had background levels of IFN-γ secretion in response to both Con A and PMA/I (Fig. 3 A, C), underscoring a defect in responsiveness to mitogens that stimulate the IFNγ signal cascade at two different levels.

Figure 3. Lymphocytes from collagen X mice display atypical responses to mitogens.

Interferon gamma (IFN-γ) ELISA was used to assess response to concanavalin A (Con A) (A, B), and to phorbol 12-myristate 13-acetate/Ionomycin (PMA/I) (C, D) by splenocytes from week-3 (A, C) and adult (B, D) wild type (WT), collagen transgenic (Col X Tg), collagen X null (Col X KO), and collagen X Tg and KO perinatal lethal mice (Col X mut, data pooled). Note, from the same pool of adult collagen X mice the Tg and KO mice either responded to Con A (Hi) or had no response to Con A (Lo). Diminished or negligible response to Con A was seen in all collagen X mice, whereas only Col X mut mice displayed an impaired response to PMA/I. (n= number of mice per group, *P< 0.05, **P< 0.001).

Mice with disrupted collagen X function do not survive parasitic challenge

The opportunistic parasite Toxoplasma gondii (T. gondii) activates the innate, humoral (B lymphocyte) and cellular (T lymphocytes) arms of the immune system (Correa et al., 2007), and was used to assess the susceptibility of the collagen X mice to infection and their ability to mount an appropriate immune response. For this purpose, the non-virulent ME49 strain of T. gondii was injected into week-5 collagen X mice with no outward disease phenotype, as well as into age-matched wild type controls. Within the first week, control and collagen X mice showed signs of sickness, manifested as ruffling of fur, hunching of back and reduced mobility; however the wild type mice were able to control the parasitic infection and survive the course of the experiment (Fig. 4). Conversely, both the collagen X Tg and KO mice developed increasingly severe diseases and succumbed to infection either within the first few weeks after challenge or gradually over 170 days (Fig. 4). Moribund mice were assessed by peritoneal lavage in conjunction with cytospin to estimate the number of infected cells at the site of challenge and both collagen X and control mice were able to clear T. gondii (data not shown). However, sections of brain, liver and lung from challenged collagen X mice showed an unusual increase in the number of parasite cysts, consistent with increased systemic parasite dissemination and deficiencies in the immune response (data not shown) (Kang et al., 2000; Johnson and Sayles, 2002; Couper et al., 2005). Moreover, challenged collagen X mice did not gain weight over the 5 month experiment and had enlarged spleens, both of which are indicative of an ongoing immune response (Argiles and Lopez-Soriano, 1999). Interestingly, histology of collagen X mouse femurs revealed hypocellular marrows with a paucity of leukocytes and an increase in erythrocytes, as well as decreases in the length and thickness of trabecular bony spicules (Fig. 5). These skeleto-hematopoietic changes were remarkably similar to the acute histological changes seen in the week-3 perinatal lethal subset of collagen X mice, which we hypothesize are dying from opportunistic infections (Gress and Jacenko, 2000; Jacenko et al., 2002). A lack of immune response regulation can cause multiple organ pathology and exuberant inflammatory responses (Bone, 1996; To et al., 2001; Evseenko et al., 2007). Therefore, the serum cytokine levels of the challenged wild type and collagen X mice were assessed for IFN-γ and IL-12, which are produced at high levels after challenge with T. gondii and can cause severe organ pathology when deregulated (Liesenfeld et al., 1996; Cai et al., 2000; Lieberman and Hunter, 2002; Gaddi and Yap, 2007). Challenged collagen X mice had significant increases in IFN-γ and decreases in IL-12 when compared to control mice (Fig. 6), which interestingly mimics the cytokine profile reported in patients with difficulty controlling an intense severe acute respiratory syndrome infection (Cameron et al., 2007).

Figure 4. Collagen X mice do not survive parasitic challenge.

Week-5 wild type (WT), collagen X transgenic (Col X Tg) and collagen X null (Col X KO) mice were challenged with a non-virulant strain of Toxoplasma gondii and monitored for survival. All challenged collagen X mice succumbed to death either rapidly, within the first few weeks of infection, or displayed a gradual demise.

Figure 5. Morphology of the femoral chondro-osseous junction reveals hypocellular marrow and reduced trabecular bone in collagen X mice after parasite challenge.

H&E stained femoral longitudinal sections from wild type (A, D) and two collagen X Tg mice that succumbed rapidly to challenge with Toxoplasma gondii (B, E and C, F). (A–C): low magnification (10X) view of the femoral chondro-osseous junction showing the curved hematoxylin-stained growth plate, eosinophilic trabecular bone (arrows), and marrow cells. (D–F): Higher magnification (40X) of marrow. Note in Tg mice a reduction of trabecular bone (B, C), a predominance of erythrocytes, and a paucity of leukocytes (E, F). Bars = 100µm.

Figure 6. Collagen X mice have increased IFN-γ and decreased IL-12 levels upon death after parasite challenge.

Interferon-gamma (IFN-γ) and IL-12 specific ELISAs were performed using serum collected from Toxoplasma gondii challenged wild type (WT), collagen transgenic (Col X Tg) and collagen X null (Col X KO) mice. (*P< 0.05, **P< 0.001)

Collagen X mice have altered serum cytokine levels

Due to the similarities between the collagen X perinatal lethal mice and the collagen X mice that succumbed to parasite challenge, e.g. similar aplastic marrow phenotype and demise likely due microbial infection (Fig. 5) (Gress and Jacenko, 2000; Jacenko et al., 2002), cytokines involved in cachexia, anemia and immune response regulation, IFN-γ, IL-12, and IL-6, were assessed for deregulation temporally in collagen X mice (Fig. 7) (Argiles and Lopez-Soriano, 1999; van der Poll and van Deventer, 1999; Chesler and Reiss, 2002; Heinrich et al., 2003; Watford et al., 2003; Malmgaard, 2004; Young, 2006). All of the collagen X mice showed altered levels of IFN-γ, IL-12, and IL-6 throughout life when compared to wild types, with increased levels measured in young and old mice. Moreover, the perinatal lethal subsets had either non-detectable or extremely high levels of cytokines (Fig. 7).

Figure 7. Collagen X mice have aberrant cytokine levels throughout life.

Interlukin-12 (IL-12) (A), interferon-gamma (IFN-γ) (B) and IL-6 (C) specific ELISAs were performed using wild type (WT), collagen X transgenic (Tg), collagen X null (KO) and perinatal lethal (mut) mouse sera. Of note, the perinatal lethal mice had either very high levels or non-detectable levels of IL-12 (A), IFN-γ (B), and IL-6 (C), which were outside the range of the plots in (B) and (C).

To further investigate cytokine changes in the collagen X mice, a cytokine protein array was performed with sera from week-3 mice. Sera pooled from collagen X Tg and KO mice with mild phenotypes and sera from two subsets of perinatal lethal mice were compared to wild types for levels of inflammatory, chemo-attractant, and matrix binding cytokines (Table 1). These data show significant increases in several cytokines involved in regulating the immune response and lymphocyte attraction, e.g. IL-4, IL-12, IL-13, cutaneous T cell-attracting chemokine (CTACK), 6Ckine and leptin (Morales et al., 1999; Vassileva et al., 1999; Myers et al., 2002; Langrish et al., 2004; Taleb et al., 2007). Of note, many of the cytokines, chemokines, and growth factors measured bind to heparan sulfate proteoglycans (HSPG) (Table 1, highlighted in blue) (Roberts et al., 1988; Ramsden and Rider, 1992; Lortat-Jacob et al., 1997; Luster, 1998; Borghesi et al., 1999; Hasan et al., 1999; Salek-Ardakani et al., 2000; Handel et al., 2005; Ellyard et al., 2007).

Table 1. Identification by serum protein array of aberrantly expressed cytokines in collagen X mice.

Sera from week-3 wild type, collagen X transgenic (Col X Tg), collagen X null (Col X KO) and collagen X Tg and KO perinatal lethal (Col X-mut) mice were analyzed for cytokine protein levels. The collagen X mouse samples were compared to WT samples to generate results shown. Significant changes were determined by Raybiotech, where + indicates ≥1.5 change over wild type levels; − indicates ≥ 0.65 change below wild type levels, Bolded cytokines were increased in all collagen X mice. Blue highlight indicates cytokines that bind to heparan sulfate proteoglycans.

Cytokine Protein Array
	Col X Tg	Col X KO	Col X mut
CTACK	+	+	+*
IL-4	+	+	+
IL12-p40/p70	+	+	+*
IL12-p70	+	+	+
IL-13	+	+	+
Leptin	+	+	+
6Ckine	+	+	-
MIP-1-alpha	-		-
VEGF		+	+*
IL-3		+
IL-10		+
G-CSF	+
MIP-2	+
TARC		+	+
RANTES		+
IL-6			-
IL-9			-
IL-17			+*
KC			-
MCP-1			+
MIP-3-beta			-
TIMP-1			+
Eotaxin
GM-CSF
IL-2
IL-5
IFN-gamma
SCF
sTNF RI
TPO
MCP-5
TNF-alpha

Note, three separately pooled collagen X perinatal lethal samples were submitted for analysis, and the * indicated difference between the three sample groups, e.g. with regards to a particular cytokine, some samples were increased while others were decreased when compared to WT.

DISCUSSION

Collagen X Tg and KO mice have always been described as having a variable disease phenotype, with subsets of mice displaying perinatal lethality within three weeks after birth and survivors having a normal life span, but showing susceptibility to skin ulcers and aggressive lymphosarcomas (Jacenko et al., 1993b; Gress and Jacenko, 2000; Jacenko et al., 2002). Additionally, flow cytometry and histologic analyses of collagen X mice have revealed altered lymphocyte profiles throughout life, as well as aberrant lymphatic organ architecture, with the most dramatic alterations observed in the perinatal lethal mice. Further, several observations imply that mice displaying perinatal lethality may be succumbing to systemic infection. For example, we have observed rampant microbial outgrowth from organs isolated from the perinatal lethal mice (Sweeney et al., manuscript in preparation), ablation of perinatal lethality with Sulfatrim®antibiotic treatment of pregnant and nursing collagen X females (unpublished observation), and recovery of mice displaying perinatal lethality within 24 hours after receiving oxytetracycline injections (unpublished observations). Taken together, these data are consistent with the possibility that immunity in all collagen X Tg and KO mice might be impaired as an outcome of altered EO; this study supports this possibility by confirming an impaired ability of the collagen X mice to mount an appropriate immune response both in vitro and in vivo.

Reduced numbers of lymphocytes (B220⁺, CD4⁺, CD8⁺, and CD4⁺/fox3P⁺) in all collagen X Tg mice, as well as in Tg and KO perinatal lethal subsets, were confirmed by flow cytometry (Fig. 1 & Fig. 2). These results suggested that the collagen X Tg and perinatal lethal mice might have a reduced ability to respond to infection, and/or an impaired capacity to regulate activated lymphocytes. In contrast, the collagen X KO mice with a mild phenotype had increased numbers of splenic lymphocytes and normal numbers of thymocytes and Tr cells (Fig. 1 & Fig. 2); this highlights some of the subtle differences between the collagen X Tg and KO mice and may explain the reduced perinatal lethality observed in the collagen X KO lines (Gress and Jacenko, 2000). Regardless, all the collagen X mice likely had a large enough lymphocyte population to respond to infection since it has been estimated that only 100–200 antigen specific T lymphocytes are needed to elicit an immune response (Blattman et al., 2002; Beverley, 2008). Therefore, it was necessary to assess the ability of the collagen X mouse lymphocytes to respond to antigenic stimuli before making conclusions on the immunologic capability of the smaller lymphocyte pool.

In vitro lymphocyte activation assays revealed defects in response to Con A from all week-3 and adult collagen X Tg and KO mice (Fig. 3). This non-specific mitogen activates T lymphocytes through the clustering of TCRs and downstream activation of the PKC/Ras intracellular signaling pathway (Stamm, 2002). Since the week-3 and adult collagen X mouse T lymphocytes had similar levels of TCRβ/CD3 complexes compared to wild type (data not shown), it is probable that the lack of stimulation was not due to decreased clustering of receptors. Next, PMA/I was used to assess the ability of collagen X mouse lymphocytes to secrete IFN-γ in response to PKC and Ca⁺⁺ stimulation. In these studies, all the collagen mice X mice were able to secrete IFN-γ after PMA/I stimulation, except the perinatal lethal subsets (Fig 3). Taken together, these findings underscored an impaired ability of lymphocytes from collagen X mice to respond appropriately to antigen stimulation in vitro; moreover, these data implicated the chondro-osseous/hematopoietic marrow environment as key in establishing peripheral immune function.

These in vitro observations were further supported by in vivo parasite challenges. Challenge of outwardly healthy collagen X Tg and KO mice with T. gondii highlighted the inability of these mice to respond appropriately to infection. All challenged collagen X mice succumbed to infection either rapidly, within the first few weeks of challenge, or displayed a gradual demise (Fig. 4). Organ histology revealed that the collagen X mice suffered from a more invasive systemic infection than wild type cohorts, however were able to clear the parasites from the site of injection as indicated by peritoneal lavage. These observations suggested that collagen X mice succumbed to the challenge due to defects in immune responses, and not from parasitemia. Additionally, morbidity within 13 to 40 days after T. gondii challenge has been described in several mouse models with known immunodeficiencies, including severe combined immunodeficiency (SCID), CD4 T cell deficient, B cell deficient, and IL10 KO mice, (Johnson, 1992; Hunter et al., 1994; Neyer et al., 1997; Kang et al., 2000; Johnson and Sayles, 2002), which substantiates the hypothesis of impaired immune function in the collagen X mice. Further, as was observed with the Con A challenges (Fig. 3), adult collagen X Tg and KO mice presented with a variable ability to mount an immune response separating the mice into two subsets: 1) mice displaying acute immune dysfunction (e.g. lack of IFN-γ secretion after Con A stimulation and rapid death following T. gondii challenge), and 2) mice displaying impaired immunity (e.g. diminished IFN-γ secretion after Con A stimulation and eventual demise following parasite challenge). Thus, the variability in disease phenotype severity of collagen X mice is not restricted to perinatal lethality at week 3, but continues to present as a variable spectrum of abnormalities in these mice throughout their life. One possible explanation for this may be the lack of growth plate closure in mice and other rodents. Thus, the biological effects of abnormal collagen X function may extend into murine adulthood, whereas in humans these effects may cease after adolescence and growth plate closure. Overall, we propose that these manifestations may ensue from a deregulated cytokine metabolism in the collagen X mice, as was suggested by aberrant serum cytokine levels in the challenged mice (Fig. 6).

In animals with an impaired ability to mount an appropriate immune response, infection may lead to cytokine over-production and an exuberant inflammatory response (Bone, 1996; To et al., 2001; Evseenko et al., 2007). Interestingly, throughout life the collagen X Tg and KO mice displayed aberrant profiles of IFN-γ, IL-12, and IL-6 cytokines, which are involved in cachexia, anemia and immune response regulation (Fig. 7) (Argiles and Lopez-Soriano, 1999; van der Poll and van Deventer, 1999; Chesler and Reiss, 2002; Heinrich et al., 2003; Watford et al., 2003; Malmgaard, 2004; Young, 2006). Moreover, these cytokines were either notably elevated or depleted in the perinatal lethal subset, which is in line with irregular immune cell signaling and cytokine toxicity (Cameron et al., 2007; Sriskandan and Altmann, 2008). Further, morphologic similarity of the chondro-osseous junction and marrow between the collagen X mice that succumbed rapidly to parasite challenge and the week-3 perinatal lethal mice was observed, and both groups of mice had aberrant cytokine levels compared to controls (Fig. 5–Fig. 7). Taken together, our data identify in the collagen X Tg and KO mice an abnormal immune response to infection, coupled with altered cytokine metabolism.

The immune dysfunctions in the collagen X mice may arise as a consequence of an altered hematopoietic microenvironment in the marrow, which is likely linked to the defects caused by disruption of collagen X function in the hypertrophic cartilage (Jacenko et al., 1993a; Jacenko et al., 1993b; Campbell et al., 2004). Likewise, these data implicate a feedback mechanism from the marrow to the peripheral immune system and back. Analysis of several human diseases provides additional precedence to this skeleto-hematopoietic-immune link. Collectively, the category of diseases manifesting both skeletal and hematopoietic changes is termed “immuno-osseous disorders” and includes a number of conditions that share phenotypic similarities with the collagen X mice. Examples of such disorders include: cartilage-hair hypoplasia (CHH), Kostmann’s syndrome, Shwachman-Diamond syndrome, Schimke dysplasia, Fanconi anemia, Diamond-Blackfan anemia, Dubowitz, Omenn and Barth syndromes, kyphomelic dysplasia, spondylo-mesomelic-acrodysplasia, and adenosine deaminase deficiency (Cederbaum et al., 1976; Schofer et al., 1991; Buchinsky et al., 1995; Corder et al., 1995; Soyer and McConnell, 1995; Duhrsen and Hossfeld, 1996; Tsukahara and Opitz, 1996; Castriota-Scanderbeg et al., 1997; Yakisan et al., 1997; Dianzani et al., 2000; Sekhar et al., 2001; Cham et al., 2002; Tischkowitz and Hodgson, 2003; Kuijpers et al., 2004; Hermanns et al., 2005; Guggenheim et al., 2006; Hubbard et al., 2006; Spencer et al., 2006; Lucke et al., 2007; Marrella et al., 2008). Although the genetic basis for a number of these disorders has been recently associated with molecules that are not obvious players in skeletogenesis or ECM establishment and maintenance, it is conceivable that similar pathways altered in the collagen X mice may be affected in some of these disorders, thus converging onto a similar disease phenotype. For example, mutations in the granulocyte colony-stimulating factor receptor are reported in Kostmann’s syndrome, yet patients present with neutropenia, frequent bacterial infections, as well as severe osteopenia (Yakisan et al., 1997; Herbst et al., 1999; Sekhar et al., 2001). Shwachman-diamond syndrome is caused by a mutation in a putative RNA metabolism gene and results in growth retardation, short stature, and bone marrow dysfunction (Boocock et al., 2003; Kuijpers et al., 2004). Finally, patients diagnosed with CHH have mutations in the RNA component of the ribonucleoprotein complex RNase MRP and present with disproportionate short stature and deficient cellular immunity (Blattman et al., 2002; Hermanns et al., 2005; Guggenheim et al., 2006; Hermanns et al., 2006), remarkably reminiscent of the collagen X murine metaphyseal dysplasia and hematopoietic defects. It may be relevant that CHH patients display altered levels of several immune mediators, as was reported in this study with the collagen X Tg and KO mice (Fig. 7 and Table 1), further connecting aberrant cytokine levels to immune defects.

Cytokine protein arrays assessed changes of 32 different immune mediators (Table 1), and confirmed altered cytokine levels in all collagen X Tg and KO mice. It is noteworthy that several of the cytokines/chemokines that are elevated in the collagen X mice are involved in immune cell development/differentiation, immune response and hematopoiesis (Patchen et al., 1991; Jacobsen et al., 1994; Eng et al., 1995; Broxmeyer, 2001). In particular, chemokines CTACK and 6Ckine assist in the movement and homing of leukocytes (Argiles and Lopez-Soriano, 1999; Handel et al., 2005), and cytokines IL-4, IL-12, and IL-13 have been implicated in enhancing hematopoiesis (Jacobsen et al., 1994; Keller et al., 1994; Eng et al., 1995). Whereas, macrophage inflammatory protein-1 alpha (MIP-1α), which is decreased in the collagen X Tg mice and perinatal lethal subsets, is implicated in inhibiting hematopoiesis (Broxmeyer et al., 1990; Bodine et al., 1991; Eaves et al., 1993). Further, it may be of relevance that many of these factors, including IL-4, IL-6, IL-12, IL-13, CTACK, and 6Ckine, bind to HSPGs (Table 1, highlighted in blue); (Patchen et al., 1991; Jacobsen et al., 1994; Keller et al., 1994; Eng et al., 1995; Broxmeyer, 2001). HSPGs have been directly implicated as key orchestrators of HSC niches in the marrow, by binding cytokines and presenting them to stromal cells and HSCs (Bruno et al., 1995; Gupta et al., 1996; Gupta et al., 1998; Borghesi et al., 1999). These findings are of particular relevance, since HSPGs are decompartmentalized in the chondro-osseous junction of the collagen X mice as a result of a disrupted collagen X network (Jacenko et al., 2001). Taken together, these data lead to a provocative possibility that there is a consistent demand for hematopoiesis in the collagen X mice due to the lack of pro-hematopoietic cytokine sequestration at the HSC niche. Additionally, this theory combines the disruption of a collagen X-containing matrix at the hypertrophic cartilage/marrow interface to altered HSPG distribution, which may suggest a potential locus for hematopoietic failure.

Hematopoietic stem cells have been shown to preferentially localize to the chondro-osseous junction where various cellular constituents and ECM components encounter to physically compartmentalize the marrow (Nilsson et al., 2001; Calvi et al., 2003; Yoshimoto et al., 2003; Zhang et al., 2003; Moore and Lemischka, 2004; Balduino et al., 2005; Sipkins et al., 2005; Taichman, 2005; Adams et al., 2006; Wilson and Trumpp, 2006). Of particular relevance, Emerson’s and Taichman’s groups demonstrated that osteoblasts are required for B-cell commitment and maturation (Taichman and Emerson, 1998; Zhu et al., 2007). We would like to extend this hypothesis to include hypertrophic chondrocytes and collagen X as players in HSC niche establishment. We propose that a function of the collagen X/HSPG network is to sequester hematopoietic cytokines and growth factors at the chondro-osseous junction. During EO, continual remodeling of trabecular bone, comprised of a hypertrophic cartilage core and osteoid, into mature secondary bone by osteoclasts may result in liberation of these HSPG bound cytokines to the local HSCs. In the collagen X mice, disruption of the lattice-like network around hypertorphic chondrocytes leads to loss of HSPGs at this site (Jacenko et al., 2001), which may impact sequestering of hematopoietic cytokines and growth factors necessary for HSC quiescence and differentiation. This scenario may explain the negative downstream effects on peripheral lymphopoiesis and immunity described in the collagen X Tg and KO mice.

EXPERIMENTAL PROCEDURES

Mouse maintenance

Mice were maintained in a virus-free barrier facility in microisolators, fed autoclaved Purina mouse chow (Animal Specialties and Provisions, LLC, Quakertown, PA) and water ad libitum. From birth, the colony was inspected daily for growth, behavioral, skeletal, or hematopoietic abnormalities as previously described (Jacenko et al., 2002). In particular, we were vigilant for decrease in size, hunching of back, lethargy, changes in mobility, and wasting, which are characteristic features of the acute perinatal-lethal phenotype in a subset of the collagen X mice around days 16–21 (Jacenko et al., 2002). Weaning at day 21 included ear punching for identification. Genotyping involved DNA isolation from a tail biopsy, phenol/chloroform purification, and genomic Southern blotting. Euthanization was by isoflurane overdose. Collagen X mouse strains used in this study included two outbred Tg lines (4.7-21D and 1.6-293D; (Campbell et al., 2004), one outbred KO line (Gress and Jacenko, 2000), as well as congenic lines that were generated by inbreeding each of these lines greater than 12 generations into the C57BL/6 mouse strain. This generated three congenic mouse strains: 1.6-293Δ -B6, 4.7-21Δ-B6, and KO-B6, which all had similar phenotypes (E. Sweeney et al., manuscript in preparation).

Parasite challenge

The ME49 strain of Toxoplasma gondii was maintained and provided by Dr. C. Hunter (University of Pennsylvania). Outbred wild type, collagen X Tg and collagen X KO five week old male and female mice were challenged with 20 cysts of ME49 delivered via an intraperitoneal injection. Throughout the course of infection, mice were monitored daily for behavioral changes, ruffled fur, lethargy, and dehydration. Moribund mice were assessed by peritoneal lavage in conjunction with cytospin to estimate the number of infected cells by microscopy as previously described (Lieberman et al., 2004). For histology and cytokine analysis at the conclusion of the experiment, hind limbs, brain, spleen, lung, liver and blood were collected.

Histology

Tissues were fixed in 4% formaldehyde/phosphate-buffered saline (PBS) (pH 7.4) at 4°C for 1 week. Fixed hind limb samples were decalcified (4% formalin, 1% sodium acetate, 10% EDTA), and all limbs and organs were then washed in deionized water, dehydrated in ascending ethanol series, cleared with Propar (Anatech, Battle Creek, MI), and paraffin-embedded. Six µm sections were stained with hematoxylin and eosin (H&E; Sigma Diagnostics, St. Louis, MO).

Lymphocyte stimulation assays

Spleens were harvested from wild type and collagen X mice at week-3 or ≥ month-5. Splenocytes were collected via grinding the spleen between two sterilized frosted slides (Fisher) over a 100mm dish and transferring cells to a centrifuge tube. Splenocytes were washed with PBS (Sigma), and the red blood cells were lysed with ACK buffer (0.15M NH₄Cl, 10mM KHCO₃, 0.1mM Na₂EDTA, pH 7.4, 1 min). Live cells were counted via hemacytometer using Trypan blue (Fisher) exclusion and were plated in 48 well plates at 3×10⁵ cells/well in RPMI medium supplemented with 10% FBS and penicillin/streptomycin (Gibco). Concanavalin A (Con A, 5µg/ml, Sigma) or Phorbol 12-myristate 13-acetate/Ionomycin (PMA/I, 50 ng/ml and 500 ng ml respectively, Sigma) were added to each well and cells were incubated at 37°C (72 hr. with Con A or 48 hr. with PMA/I). Supernatants were collected and stored at −20°C, and 200µl was used in IFN-γ ELISA, as described below.

ELISA

ELISAs were performed as described in the Short Protocols in Immunology (Coligan, 2005). Briefly, blood was collected from euthanized mice via heart puncture, allowed to clot at room temperature and then serum was isolated via centrifugation (6000 RPM, 10min 4°C). Serum cytokines were measured by coating ELISA plates (Thermo Scientific) with 5ug/ml capture antibody: IFN-γ, IL-12, and IL-6 antibodies (clones R4-6A2, C15.6, and 20F3 respectively) and incubating with 50µl serum. Cytokine specific detection antibodies (1ug/ml, BD Pharmingen) were added and incubated at room temperature for one hour. Strepavidin-alkaline phosphatase conjugate (1:2000, BD Pharmingen) and pNPP (100µl, in solvent, 1M diethanolamine, 0.5 mM MgCl₂, pH 9.8, Sigma, St. Louis, MO) were used to quantify cytokines. An ELISA plate reader (405nm, Biorad Benchmark Plus) was used to generate sample curves, which were compared with IFN-γ, IL-12, and IL-6 standard curves created with recombinant proteins (BD Pharmingen). Results were plotted in Excel and two-tailed, two sample unequal variance T-tests were used to establish significance.

Flow cytometry

Flow cytometry was performed as described in Short Protocols in Immunology (Coligan, 2005). Briefly, spleen and thymus tissues were harvested from wild type and collagen X mice at week-3 and single cell suspensions were made using a Tenbroeck Tissue Grinder (Wheaton, Millville, NJ) as previously described (Jacenko et al., 2002). Erythrocytes were lysed with ACK buffer (1 min, as detailed above) and cells re-suspended in flow buffer (PBS pH 7.2, 0.5% BSA, 2mM EDTA). Cells were labeled in flow buffer with specific antibodies (0.2µg/10⁶ cells): anti-TCRβ FITC, anti-CD3 APC, anti-CD4/L3T4, anti-CD8a/Ly-2, anti- CD45R/B220, and isotype controls (BD PharMingen). When necessary, cells were then stained with FITC polyclonal anti-rat IgG (BD PharMingen) as secondary antibody (0.1µg/10⁶ cells). For live cell count analysis, propidium iodide (PI; BD PharMingen; 10 µl at 50 µg/ml) was added for dead cell exclusion and the FACSCalibur was used with CELL Quest 3.1 (Becton Dickinson, Mansfield, MA) for data acquisition and analysis. For T regulatory cell analysis, splenocytes were labeled as per the MACS Anti-FoxP3 kit using anti-CD4 Percp (BD Pharmingen) (0.2µg/10⁶ cells) and antibodies provided. The BD Canto was used with DiVa software for data acquisition and FlowJo software for data analysis.

Protein array

Blood was collected from euthanized mice via heart puncture and serum was isolated (as detailed above). The Mouse Cytokine Antibody Array System II & 2.1 cytokine antibody array from Raybiotech (Norcross GA) was chosen to assess 32 serum cytokines. Eight pooled serum samples (three mice per sample), including two wild type controls, two collagen X transgenic, one collagen X null and three perinatal lethal mouse samples were sent to Raybiotech. The three different collagen X perinatal lethal mutant mouse samples were selected based on high or low IL-12 levels as determined by IL-12 ELISA prior to shipping. Analysis was performed by Raybiotech and included averaging duplicates, subtracting out background and normalizing data based on positive controls (secondary antibody only) to accurately compare cytokine levels between different sample filters.