MHC Class II Transactivator Is an In Vivo Regulator of Osteoclast Differentiation and Bone Homeostasis Co-opted From Adaptive Immunity

Authors


  • For a Commentary on this article, please see Nakamura (J Bone Miner Res. 2014;29:287–289. DOI: 10.1002/jbmr.2161).

ABSTRACT

The molecular networks controlling bone homeostasis are not fully understood. The common evolution of bone and adaptive immunity encourages the investigation of shared regulatory circuits. MHC Class II Transactivator (CIITA) is a master transcriptional co-activator believed to be exclusively dedicated for antigen presentation. CIITA is expressed in osteoclast precursors, and its expression is accentuated in osteoporotic mice. We thus asked whether CIITA plays a role in bone biology. To this aim, we fully characterized the bone phenotype of two mouse models of CIITA overexpression, respectively systemic and restricted to the monocyte-osteoclast lineage. Both CIITA-overexpressing mouse models revealed severe spontaneous osteoporosis, as assessed by micro-computed tomography and histomorphometry, associated with increased osteoclast numbers and enhanced in vivo bone resorption, whereas osteoblast numbers and in vivo bone-forming activity were unaffected. To understand the underlying cellular and molecular bases, we investigated ex vivo the differentiation of mutant bone marrow monocytes into osteoclasts and immune effectors, as well as osteoclastogenic signaling pathways. CIITA-overexpressing monocytes differentiated normally into effector macrophages or dendritic cells but showed enhanced osteoclastogenesis, whereas CIITA ablation suppressed osteoclast differentiation. Increased c-fms and receptor activator of NF-κB (RANK) signaling underlay enhanced osteoclast differentiation from CIITA-overexpressing precursors. Moreover, by extending selected phenotypic and cellular analyses to additional genetic mouse models, namely MHC Class II deficient mice and a transgenic mouse line lacking a specific CIITA promoter and re-expressing CIITA in the thymus, we excluded MHC Class II expression and T cells from contributing to the observed skeletal phenotype. Altogether, our study provides compelling genetic evidence that CIITA, the molecular switch of antigen presentation, plays a novel, unexpected function in skeletal homeostasis, independent of MHC Class II expression and T cells, by exerting a selective and intrinsic control of osteoclast differentiation and bone resorption in vivo. © 2014 American Society for Bone and Mineral Research.

Introduction

The immune system and the skeleton have been shown to be tightly interconnected and to share a number of regulatory signals, both at the intracellular and intercellular level.[1] A paradigmatic example is the dual role of tumor necrosis factor (TNF)-related activation-induced cytokine (TRANCE)/receptor activator of NF-κB ligand (RANKL) as a regulator of dendritic cell (DC) and osteoclast (OC) activity.[2, 3] Additional exemplar cases include the function of immunoreceptor tyrosine-based activation motif (ITAM)-dependent costimulatory signals activated by immunoreceptors in OC differentiation,[4, 5] and the regulation of the hematopoietic niche by osteoblasts (OB).[6, 7]

A distinctive feature of adaptive immunity is its capacity to present foreign antigens (Ag) in the context of major histocompatibility complex (MHC) molecules for recognition by T cells. The MHC Class II (MHC-II) pathway presents peptides derived from proteins that accessed the endosomal compartment for recognition by CD4+ T cells. The expression of MHC-II is tightly regulated, being constitutive in professional Ag-presenting cells (including DC, B cells, macrophages, and thymic epithelial cells), and inducible in nonprofessional Ag-presenting cells (eg, endothelial cells, fibroblasts, epithelial cells).[8] The MHC Class II Transactivator (CIITA) is a non-DNA-binding transcriptional co-activator that is essential for the expression of genes encoding MHC-II molecules and the accessory Invariant chain (Ii), which ensure intracellular transport and surface presentation of antigenic peptides. Because CIITA is required for the development and activation of CD4+ T cells, mutations that inactivate it cause the Bare Lymphocyte Syndrome (BLS), a severe inherited immunodeficiency resulting in recurrent infections and death in early childhood.[9, 10] The variegated expression of MHC-II in different cells and over time is accounted for by the selective usage of distinct promoters driving constitutive and inducible CIITA transcription in myeloid and non-myeloid cells.[11] Unlike most mammalian transcription factors, which exert pleiotropic functions, CIITA is believed to be dedicated for Ag presentation. Long suggested by the absence of other overt phenotypes in BLS, this was conclusively demonstrated by genome-wide analyses that identified only a small number of target genes in addition to those constituting the MHC module, all implicated directly or indirectly in Ag presentation.[12]

Bone-wasting diseases are mediated by alterations affecting the balanced remodeling activities of bone-forming OB and bone-resorbing OC.[13] In particular, estrogen deficiency, the main causative factor for involutional osteoporosis, causes bone loss primarily through increased OC activity, not compensated for by a corresponding increase in bone formation. Excessive resorption largely stems from increased OC differentiation from monocytic precursors, through a mechanism that is not fully understood.[14-16]

The expression of CIITA in the OC lineage is implied by the recent report that human in vitro-generated OC express MHC-II and are able to present allogeneic Ag to T cells.[17] However, whether the MHC-II pathway in turn regulates OC function and bone homeostasis is unknown. An association between CIITA expression and osteoclastogenesis (OCgenesis) has been found in the mouse model of osteoporosis induced by surgical menopause (ovariectomy), a condition known to increase OC differentiation and activity, in which we observed increased levels of CIITA transcripts in OC precursors.[18] This rationale prompted us to test if CIITA and the MHC-II pathway play a direct role in bone homeostasis.

Adopting mouse models of systemic and lineage-restricted CIITA overexpression, we here provide genetic evidence for a novel role of CIITA as a critical regulator of adult skeletal homeostasis in vivo through the selective stimulation of OC differentiation. Our findings also establish two novel genetic models of spontaneous, early onset and severe osteoporosis of obvious biomedical and biotechnological interest.

Materials and Methods

Mice

All animal procedures were approved by the Institutional Animal Care and Use Committee of the San Raffaele Scientific Institute, Milano, Italy (IACUC #291, 377, 539), where all animal procedures were conducted. CIITA pIV−/−, CIITA Tg, MHC-II−/− (MC42), and pIV−/− K14 CIITA Tg mice were generated on a C57Bl/6 background as described.[19]

Cell cultures and osteoclastogenesis

Bone marrow monocytes (BMM) were purified in whole bone marrow (BM) cultures in α-MEM with 100 ng/mL macrophage colony-stimulating factor (M-CSF). M-CSF-producing cells were kindly provided by Prof Takeshita (National Center for Geriatrics and Gerontology, Obu, Japan). For CIITA mRNA expression, DC were generated by culturing BMM with GM-CSF and IL-4, both at 25 ng/mL, for 5 days. Spleen and BM B cells were purified respectively with immunomagnetic negative and positive (B220 microbeads) selection (Miltenyi, Teterow, Germany). BM stromal cells (SC) were purified from whole BM cells by in vitro propagation for 2 weeks in D-MEM (Life Technologies, Carlsbad, CA, USA) supplemented with 100 U/mL penicillin, 100 µg/mL streptomycin, and 1.5 mM L-glutamine (VWR, Radnor, PA, USA), in presence of high serum (20% heat-inactivated fetal calf serum [FCS], EuroClone, Pero, Italy). For OCgenesis, 104 BMM were plated in 200 µL/well in 96-well plates with 10 ng/mL M-CSF and 100 ng/mL RANKL. For OCgenic co-cultures, SC and BMM from mutant mice and wild-type (WT) controls were co-cultured (105 SC and 2 × 105 BMM per well) in 6-well plates, treated with 10−8 M 1,25 (OH)2 vitamin D3 (Sigma, St. Louis, MO, USA), and the yield of multinucleated tartrate-resistant acid phosphatase (TRAP)+ OC-like cells assayed after 2 weeks. After fixation (10 minutes 3.7% formaldehyde), cells were stained for TRAP activity using the Leukocyte Acid Phosphatase Kit (Sigma). For in vitro resorption studies, purified BMM were differentiated into OCs on dentine discs (Immunodiagnostic Systems, Boldon, UK). After fixation and TRAP staining for OC identification, cells were detached with NaOH 1M and resorption pits stained with toluidine blu. Resorption areas were measured by the ImageJ software.

Macrophage polarization and DC differentiation

Polarized macrophages were generated as described.[20] Briefly, after 7 days of culture in α-MEM (Life Technologies) containing 10% FCS (Lonza, Basel, Switzerland) in the presence of 100 ng/mL M-CSF, BMM from C57BL/6 female mice were differentiated in inflammatory (M1) or immunoregulatory (M2) macrophages, culturing them for 2 days with 50 ng/mL recombinant murine IFNγ (Peprotech, Rocky Hill, NJ, USA), or for 4 days with 10 ng/mL recombinant murine IL-10 (R&D Systems, Minneapolis, MN, USA), respectively. Macrophage polarization was verified by flow cytometry (see below). Cytokine secretion was assessed by ELISA (see below).

BM-derived DCs were prepared from flushed tibias from 4 WT and CIITA pIV−/− mice by in vitro propagation for 1 week in Iscove's medium (Life Technologies) supplemented with 100 U/mL penicillin, 100 µg/mL streptomycin and 1.5 mM L-glutamine (VWR), 10% heat-inactivated FCS (EuroClone), and recombinant mouse GM-CSF and IL-4 (both 25 µg/mL, R&D Systems). DC differentiation was verified by flow cytometry after 48 hours of treatment with LPS (1 mg/mL, Sigma-Aldrich).

FACS analyses

Macrophage polarization was verified on a FACS CANTO (BD Pharmingen, San Diego, CA, USA) followed by analysis with the FlowJo software (Treestar Inc., Ashland, OR, USA) after staining with specific fluorochrome-conjugated antibodies against the myelomonocytic marker CD11b (APC-conjugated, clone M1-70), the MHC-II molecule I-Ab (FITC-conjugated, clone C3H.SW), and the co-activating molecules CD86 (FITC-conjugated, clone GL1), all from BD Pharmingen. DC differentiation was verified on a FACS Canto TM (Becton Dickinson, Franklin Lakes, NJ, USA) upon labeling with the following PE- or FITC-conjugated antibodies (BD Pharmingen): CD11c (clone HL3), CD86 (clone GL1), CD80 (clone 16-10A1), MHC I (clone 25-g-17), or MHC class II (clone AF6-88.5). Before analysis, cells were washed in PBS 7-Amino-actinomycin D (7-AAD) added for dead cell exclusion. Data were analyzed using DIVA Software (Becton Dickinson). Expression of surface RANK was determined upon labeling with mouse-RANK affinity-purified goat IgG and anti-goat FITC-conjugated Ab (R&D Systems). For cell proliferation, 2 × 106 BMM and pre-OC were stained (8 minutes RT) with 1 µM carboxy-fluorescein succinimidyl ester (CFSE), and dilution from baseline measured 48 hours later by FACS.

ELISA

TNFα and IL-10 were assayed in culture supernatants (DuoSet, R&D). Serum CTX and osteocalcin were measured with the RatLaps EIA (Immunodiagnostic Systems) and the Mouse Osteocalcin EIA kit (Biomedical Technologies Inc., Stoughton, MA, USA), as per manufacturers' instructions.

Semiquantitative and real time RT-PCR

Total RNA was extracted with TRIZOL (Invitrogen), suspended in DEPC water, and 1 µg reverse-transcribed (RT) using Superscript First-Strand Synthesis System or SuperScript VILO cDNA Synthesis Kit respectively for semiquantitative (CIITA and β-actin) or quantitative assays (Invitrogen). RT products were diluted 1:2 or 1:5, respectively, and 1 µL amplified with 500 nM primers. Calcitonin receptor (CTR) and TRAP were quantified using mouse probe #15 for CTR and #60 for TRAP (Universal Probe Library Assay Design Centre, Hoffman-La Roche, Basel, Switzerland), with 58°C AT. RT products were diluted 1:4, and 1 µL cDNA used for qPCR (Light Cycler 480 and dedicated amplification mix, Roche). PCR protocol details and primer sequences are described as follows.

CIITA total: FW 5'-TGGGATCTTCCAGCGGAAGC-3'; RW 5'-ACAACAGGGCTGTGACTATAGC-3' (AT = 61°C); β-actin: FW 5'-GGCATCCTGAAGT-3'; RW 5'-CGGATGTCAACGTCACACTT-3' (AT = 55°C); CIITA type I: FW 5'-CAGGGACCATGGAGACCATAGT-3'; RW 5'-CAGGTAGCTGCCCTCTGGAG-3' (AT = 67°C); CIITA type III: FW 5'-GGTTCCTGGCCCTTCTGG-3'; RW 5'-ATCCATGGTGGCACACAGACT-3' (AT = 67°C); CIITA type IV: FW 5'-CAGCACTCAGAAGCACGGG-3'; RW 5'-ATC CATGGTGGCACACAGACT-3' (AT = 61°C); TATA-binding protein (TBP): FW 5'-ATGCTGAATATAATCCCAAGCGA-3'; RW 5'-GAAAATCAACGCAGTTGTCCG-3' (AT = 61°C). CTR: FW 5'-GGTTCCTTCTCGTGAACAGGT-3'; RW 5'-AGAACTGGAGTTGGGCTCAC-3'; TRAP: FW 5'-CGTCTCTGCACAGATTGCAT-3'; RW 5'-AAGCGCAAACGGTAGTAAGG-3'. Total CIITA mRNA upon RNAi was quantified using mouse probe #110 for CIITA and #107 for the housekeeping TBP (Universal Probe Library Assay Design Centre, Hoffman-La Roche), with the following primers: CIITA: FW 5'-GATGTGGAAGACCTGGATCG-3'; RW 5'-TGCATCTTCTGAGGGGTTTC-3'; TBP: FW 5'-GGCGGTTTGGCTAGGTTT-3'; RW 5'-GGGTTATCTTCACACACCATGA-3'. Ten µl reaction mix included 1X SYBR-green I mix (Hoffman-La Roche) and 4 mM MgCl2. Amplifications were performed in Light Cycler 1.0 (Roche) in triplicate as follows: hot start (denaturation), 95°C for 10 minutes; amplification (3 steps, 45 cycles): denaturation, 95°C for 10 seconds; annealing temperature (AT, see above) for 0 seconds; extension, 72°C for 20 seconds; melting curve: 95°C for 10 minutes; cooling to 30°C.

RNA interference

Lentiviral vectors expressing anti-CIITA shRNAs were packaged with Mission shRNAs (449 and 452, Sigma), pMD2-VSV-G, pMDLg/pRRE, and pRSV-Rev in HEK 293T cells for 12 hours. Then, medium was replaced and 1 mM Na-butirrate added. Forty-eight hours later, cell supernatants were collected, centrifuged, filtered, and administered to BMM with polybrene (8 µg/mL) for 24 hours. Transduced BMM were selected for 1 week with 2 µg/mL puromycin.

Immunofluorescence

Purified BMM, pre-OC (3 days after OCgenic stimulation) and OC were fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100. Nuclei were stained with Hoechst and actin ring was stained with Alexa Fluor 594 Phalloidine (Invitrogen). Analyses were performed under Olympus 1 × 70 with Deltavision RT Deconvolution system microscope, CCD camera, 20× dry and 40× oil objectives.

Histology and histomorphometry

Nondecalcified bone sections were prepared from excised tibias upon fixation in 4% paraformaldehyde, dehydration in 70% ethanol, and cryopreservation in dry ice, as described.[21-23] Briefly, 6-µm sections were obtained every 200 µm using a Bright Cryostat (–30°C) with high-profile blade (pfm medical ag, Cologne, Germany), collected on a glue tape (Sigma), transferred on adhesive-coated slides, exposed to UV light 3', and stored overnight at –20°C. Sections for histomorphometry were stained with hematoxylin and eosin (H&E) and TRAP. Histomorphometric measurements were carried out by a blinded operator according to the ASBMR guidelines using the Alexa Image Pro Plus 2 software (Media Cybernetics, Bethesda, MA, USA). To determine mineral apposition rate (MAR), we administered calcein (1 µg/g body weight) ip 10 and 3 days before euthanization, and the mean distance between labels (20 measures per section, 6 sections per mouse) determined using an Axioplan2 microscope HBO-100W/2 light source with EY-455 excitation filter for fluorescent labels, and divided by the labeling period. For bone formation rate (BFR), MAR values were multiplied by mineralizing perimeter/bone perimeter and expressed on a yearly basis.

Protein extraction and immunoblotting

Purified BMM and pre-OC were lysed in RIPA buffer. For caspase 3 immunoblotting, cells were lysed with no further treatment; for OCgenic signaling, BMM and pre-OCs were starved for 6 or 3 hours, respectively, in α-MEM without FBS and antibiotics, and treated with 100 ng/mL M-CSF for up to 20 minutes, or 100 ng/mL RANKL for up to 60 minutes, before lysis. Twenty to 40 µg proteins were loaded on a 10% polyacrilamide gel, transferred on PVDF membrane, and then blotted with the following Ab: p44/42 MAP kinase (9102); phospho-p44/42 MAP kinase (9101); SAPK/JNK (9258); phospho-SAPK/JNK (9255); AKT (4685); phospho-AKT (9275); phospho-IkBα (2859, Cell Signaling Technology, Danvers, MA, USA); NFATc1 (sc-7294, Santa Cruz Biotechnology, Santa Cruz, CA, USA); β-actin (A1978, Sigma); caspase 3 (9662, Cell Signaling Technology); α-tubulin (DM1A, Sigma-Aldrich).

pQCT

Trabecular and cortical BMD were measured in vivo and ex vivo by XCT Research SA+ (Stratec, Birkenfeld, Germany), as described.[24]

Micro-CT scans and microarchitectural analyses

Tibias were assayed in a cone-beam micro-CT scanner (Skyscan 1072, Bruker, Brussels, Belgium), as described.[25] Scanner settings: 50kVp, 200µA, rotation step 0.45°, rotation over 185°, 1 mm Al filter, scan resolution 6 µm/pixel. Cross-section images: filtered back-projection (NRecon, V 1.4.4, Skyscan), 1024 × 1024 pixels, 6 µm isotropic voxel, interslice distance 1 pixel, 256 gray level. A trabecular VOI 1.5 mm high was extracted (CT Analyser V 1.8.0.5, Skyscan) 100 µm under the growth plate. Gray-level images were binarized through a uniform threshold.[26-28] Morphometric parameters (CT Analyser): bone volume fraction [BV/TV, marching cubes method27], trabecular thickness and separation (Tb.Th., Tb.Sp., direct methods), trabecular number (Tb.N.).[28]

Statistical analyses

Bone parameters, OC yields, cytokine secretion, and expression levels were compared through a two-tailed Student's t test. Mice group sizes were determined as the minimum required to unveil 2 STD differences, with 90% potency and 5% type I error probability.

Results

To address if CIITA plays a role in bone biology, we employed genetically modified mouse models that overexpress CIITA (CIITA Tg), which have no overt immune phenotype.[29, 30] We also used mice that selectively lack CIITA promoter IV (pIV), which is essential for MHC-II expression in thymic epithelial cells and thus for positive selection of CD4+ T cells.[19, 31] We found that CIITA pIV−/− mice naturally and selectively overexpress CIITA in the monocytic lineage in the bone marrow (BM) (Fig. 1). Interestingly, CIITA was not overexpressed in BM-derived DC, splenic monocytes, or B lymphocytes from BM or spleen (Fig. 1B). Isotype expression analysis showed that in pIV−/− BM monocytes (BMM) and pre-OC CIITA is overexpressed through increased usage of pI (Fig. 1C), a promoter known to be sufficient to ensure basal and induced CIITA/MHC-II expression in the macrophage lineage.[32] CIITA Tg and pIV−/− mice thus constitute two valuable models in which CIITA overexpression is respectively either systemic or restricted to the monocyte lineage in the BM, which includes OC precursors.

Figure 1.

CIITA pIV−/− mice overexpress CIITA in the bone marrow (BM) monocyte lineage. (A) Nonquantitative RT-PCR for CIITA mRNA expression in wild type (WT), CIITA pIV−/−, and CIITA Tg BMM. One representative experiment out of three is shown. (B) Quantitative real-time RT-PCR for total CIITA mRNA in BM-derived monocytes (mono), dendritic cells (DC), and B cells, and spleen-derived monocytes and B cells from CIITA pIV−/− mice and WT control littermates (n = 5 mice per group). Data are normalized by the levels detected in WT counterparts. One representative experiment out of three is shown. (C) Quantitative real-time RT-PCR for different CIITA transcripts driven by promoters I, III, and IV in BMM purified from WT and CIITA pIV−/− mice and stimulated for 3 days with 10 ng/mL M-CSF and 100 ng/mL RANKL (pre-OCs). Data show mean ± SD of two independent experiments, each pooling cells from three different individuals. *p < 0.05.

CIITA-overexpressing mouse models develop severe osteoporosis

We first analyzed whether systemic and conditional overexpression of CIITA alters bone homeostasis in vivo in growing and adult mice. First, we assessed bone mineral density (BMD) in vivo longitudinally up to 12 weeks of age by peripheral quantitative computed tomography (p-QCT), which revealed no difference in trabecular and cortical BMD between CIITA-overexpressing mice and WT control littermates (Fig. 2A). However, 24-week-old CIITA pIV−/− mice showed a significant defect in trabecular BMD in both genders (Fig. 2A, right panel). To obtain 3D analyses of the microarchitecture, micro-CT scans were performed on excised tibias from CIITA-overexpressing mice and WT littermates. Both CIITA Tg and pIV−/− mice showed a dramatic trabecular bone defect, with trabecular number and bone volume fraction decreased by approximately two thirds (Fig. 2B, C). As a result, trabecular separation was approximately double in both genotypes (Fig. 2C). The skeletal defects observed in CIITA Tg and pIV−/− mice were comparable (Fig. 2C). The data demonstrate that CIITA-overexpressing mice display normal skeletal maturation but acquire profound trabecular defects in the early adulthood in both genders.

Figure 2.

Severe trabecular defect in CIITA-overexpressing mice. (A) Quantitative assessment of bone mineral density (BMD) in growing and adult CIITA pIV−/− mice by p-QCT. Left and middle panels, respectively, show trabecular and cortical BMD in growing (6- and 12-week-old) wild-type (WT) and mutant mice. The right panel shows trabecular BMD in excised tibias from 6-month-old WT and CIITA pIV−/− mice of both genders (n = 6 mice per group). (B, C) Qualitative and quantitative assessment of bone microarchitecture by micro-CT in tibias excised from adult (5-month-old) WT, CIITA Tg, and CIITA pIV−/− mice (n = 6 mice per group). (B) Representative 3D reconstructions of tibial epiphyses. (C) Average quantifications of trabecular bone volume/tissue volume (BV/TV), trabecular thickness (Tb.Th), number (Tb.N.), and separation (Tb.Sp.). *p < 0.05 compared with WT controls. Data from WT of the two lines, statistically identical, were pooled into single control bars to visualize the statistical analyses revealing no difference between trabecular BV/TV, number, and separation of the two mutants.

CIITA controls bone homeostasis by selectively regulating OC number and bone resorption in vivo

We next performed bone histomorphometry on histological sections of excised, cryopreserved tibias to visualize and quantify bone cell numbers and activity in vivo. These analyses confirmed the remarkable reduction of the trabecular bone network already observed by micro-CT (Fig. 3A, B). In CIITA-overexpressing mice, OB numbers did not differ significantly from those measured in sections from WT littermates (Fig. 3B). In contrast, sections from both CIITA Tg and pIV−/− mice showed substantial and comparable increases in OC numbers and in eroded bone, with normal OC size (Fig. 3B, C). To further ascertain the cellular bases of the skeletal phenotype observed, we then assessed bone remodeling in vivo. Both CIITA Tg and pIV−/− mice showed dramatically increased circulating levels of C-terminal telopeptides of collagen type I, demonstrating increased bone resorption (Fig. 3D), but normal mineral apposition and bone formation rates compared with WT controls, as assessed upon double calcein labeling in vivo (Fig. 3E). Moreover, circulating levels of osteocalcin, an indirect in vivo marker of bone formation, in both CIITA Tg and pIV−/− mice were similar to WT controls (Supplemental Fig. S1). Together, the data indicate that the skeletal spontaneous profound osteoporotic phenotype displayed by both CIITA-overexpressing mouse models is largely, if not entirely, the result of increased OC activity, with no tentative compensation by OB. The loss of the paradigmatic coupled activity of OC and OB is likely to explain the severity of the phenotype observed.

Figure 3.

Increased OC numbers and bone resorption in CIITA-overexpressing mice. (A) Hematoxylin and eosin (H&E) staining of histological sections of excised, cryopreserved tibias from WT, CIITA Tg, and CIITA pIV−/− mice (n = 6 mice/group). Scale bars = 100 µm. Representative H&E-stained images. (B) Histomorphometric analyses of bone parameters in vivo (n = 6 mice/group). Average quantifications in the sections analyzed include: quantification of trabecular bone volume/total volume (BV/TV); OB number/bone surface; OC number/bone surface; eroded perimeter/bone perimeter (E.Pm/B.Pm); OC perimeter/bone perimeter (Oc.Pm/B.Pm). As in Fig. 1, data from WT groups of the two mutant models, statistically identical, were pooled into single control bars. *p < 0.05. (C) TRAP staining of histological sections of excised, cryopreserved tibias from WT, CIITA Tg, and CIITA pIV−/− mice (n = 6 mice/group). Scale bars = 100 µm. Arrowheads point to multinucleated, TRAP+ OC. (D, E) In vivo assessment of bone remodeling (n = 6 mice per group). (D) Circulating levels of C-terminal telopeptides of collagen type I (CTx) as assessed by ELISA. *p < 0.05. (E) Mineral apposition rate (MAR) and bone formation rate (BFR) as assessed by double calcein labeling in vivo, 10 and 3 days before death. Representative images (left) and average quantifications (right) are shown. As above, WT data of the two lines were found to be identical and pooled into single control bars.

CIITA pIV deficiency causes osteoporosis independent of T cells

T cells have been shown to play critical roles in bone pathophysiology.[33] Owing to absent MHC-II in thymic epithelial cells, CIITA pIV−/− mice lack CD4+ T cells, whereas CIITA Tg exhibit normal thymic CD4+ T-cell development. We thus hypothesized that CIITA exerts an effect on bone metabolism independent of T cells. To conclusively address this point, we employed an additional CIITA pIV−/− mouse line, pIV−/− K14 CIITA Tg mice, which express CIITA in the thymic epithelial cortex, hence restoring a normal CD4+ T-cell repertoire.[34] Micro-CT and histological analyses revealed a dramatic trabecular defect in these mice, comparable to that observed in CIITA Tg and pIV−/− mice (Fig. 4). The data demonstrate the T-cell independence of the bone-wasting effect of CIITA.

Figure 4.

Severe trabecular defect in pIV−/− K14 CIITA Tg mice. (A) Micro-CT analysis of proximal tibial epiphyses from pIV−/− K14 CIITA Tg mice and wild-type (WT) controls (n = 5 per group). Average quantifications (±SD) of trabecular bone volume/tissue volume (BV/TV), thickness (Tb.Th), number (Tb.N.), and separation (Tb.Sp.). *p < 0.05. (B) Hematoxylin and eosin (H&E) staining of histological sections of excised, cryopreserved tibias from WT and pIV−/− K14 CIITA Tg mice. Representative images are shown.

CIITA is a positive, intrinsic regulator of OC differentiation

In search for a direct effect of CIITA on bone cells, we next hypothesized that CIITA regulates the OCgenic potential of OC precursors. To address this issue, we performed ex vivo OCgenic assays on BMM purified from overexpressing CIITA mutants. When BMM from CIITA pIV−/− and CIITA Tg mice were stimulated with M-CSF and RANKL, the yield of giant multinucleated, TRAP+, OC-like cells was greatly increased (Fig. 5A, B) compared with parallel cultures from BMM purified from WT littermates. Gene expression of the specific OC markers TRAP and calcitonin receptor (CTR) were similarly increased, on a total RNA basis, proportional to increased OC yields (Supplemental Fig. S2A, B). Similarly, overall resorption on dentin was increased, in proportion with increased OC yields (Supplemental Fig. 2C, D). To establish if CIITA plays an intrinsic role in OCgenesis, we silenced endogenous CIITA by RNA interference (RNAi) in WT BMM before inducing OC differentiation. Reduced CIITA mRNA expression strongly and dose-dependently decreased OCgenesis (Fig. 5C).

Figure 5.

CIITA is a positive regulator of osteoclastogenesis. (A, B) OC differentiation assays performed with BMM purified from CIITA pIV−/− (A) and CIITA Tg (B). Cells were differentiated in the presence of 10 ng/mL M-CSF and 100 ng/mL RANKL in 96-well plates, fixed, and stained for TRAP activity in triplicate wells at different time points. Top panels show representative fields after 6 days of culture. Scale bars = 50 µm. Histograms quantify the average numbers of multinucleated TRAP+ cells (OC) at the indicated times. Data represent mean ± SD derived from three independent experiments, each using BMM pooled from three littermates. (C) OCgenesis upon treatment with lentiviral RNA interference (RNAi) directed against CIITA in purified BMM. Two distinct interfering vectors were utilized, and results were compared with an unrelated control vector. The bottom left histogram shows CIITA mRNA silencing as assessed by quantitative real-time RT-PCR. Cells were then stimulated with M-CSF and RANKL as above, and OC yields were determined by staining for TRAP activity 7 days post-stimulation. The bottom right histogram shows average OC counts derived from three parallel experiments (mean ± SD). The top panels show representative images from OCgenic cultures. Scale bar = 50 µm. *p < 0.05.

To gauge a possible contribution by stromal cells (SC) to increased OCgenesis, we purified BM SC from CIITA Tg and CIITA pIV−/− mice and from WT littermates, and assayed them for CIITA expression and OCgenic activity in standard BMM/SC OCgenic co-culture experiments. As shown in Supplemental Fig. S3, BM SCs from WT and mutant mice expressed virtually undetectable levels of CIITA, consistent with their mesenchymal origin, and showed comparable OCgenic activity. Taken together, these results demonstrate that CIITA is an intrinsic, positive regulator of OCgenesis within the BM monocyte lineage.

CIITA overexpression does not alter macrophage polarization and DC differentiation

Given the established immune function of CIITA, we then asked if its overexpression alters the differentiation of monocytic precursors into specialized immune effectors, namely macrophages and DC, or if it rather affects selectively OCgenesis. To test macrophage polarization, we analyzed cell surface marker expression and cytokine secretion on pIV−/− BMM upon differentiation into inflammatory (M1) and immunoregulatory (M2) macrophages as achieved by IFNγ and IL-10 stimulation, respectively. Flow cytometry analyses revealed that pIV−/− BMM and polarized M1 and M2 macrophages expressed the myeloid marker CD11b at levels comparable to those detected in WT controls (Fig. 6A). Attesting to normal polarization, the surface expression of MHC-II and of the co-stimulatory molecule CD86 were similarly upregulated in WT and mutant M1 and M2 macrophages (Fig. 6A). Moreover, ELISA demonstrated remarkable production of TNFα and IL-10, respectively, upon polarization in presence of IFNγ (M1) and IL-10 (M2 macrophages), with no difference between mutant and WT cells (Fig. 6B). We then assessed DC differentiation from CIITA-overexpressing BMM by surface expression of co-stimulatory molecules, CD80 and CD86, and of MHC-I and MHC-II. Apart from the expected increased basal expression of MHC-II, the canonical transcriptional target of CIITA, which was proportionally amplified also upon acquisition of the DC phenotype, CIITA pIV−/− BMM showed normal DC differentiation (Fig. 6C). Altogether, the data demonstrate that CIITA overexpression does not alter differentiation into macrophages and DC, while selectively affecting OCgenesis.

Figure 6.

CIITA overexpression does not affect macrophage polarization and DC differentiation. (A) Flow cytometric analyses of the surface expression of CD11b, MHC-II (I-Ab), and CD86 in BMM and in inflammatory (M1) and immunoregulatory (M2) macrophages differentiated from CIITA pIV−/− and wild-type (WT) BMM, respectively, in the presence of 50 ng/mL IFNγ for 2 days or M-CSF and IL-10 (both 10 ng/mL) for 4 days. Histograms from one representative of three experiments are shown. Each panel shows the expression of the indicated marker (tinted line) and appropriate isotype control (black line). Cells numbers are expressed as % of maximum number acquired. (B) TNFα and IL-10 secretion as assessed by ELISA on supernatants of M1 and M2 macrophages. Mean ± SD from three independent experiments are shown. (C) Flow cytometric analyses of surface expression of CD80, CD86, MHC-I, and MHC-II in BM-derived DCs (prepared as in Materials and Methods) treated for 48 hours with LPS (1 mg/mL) or left untreated. Graphs show the proportion of cells double-positive for the DC marker CD11c and each of the indicated markers. *p < 0.05; **p < 0.005; ***p < 0.001.

Normal bone architecture and osteoclastogenesis in MHC Class II deficient mice

To investigate the mechanism whereby CIITA controls OC differentiation and bone homeostasis, we first analyzed mice lacking MHC-II, CIITA's canonical transcriptional target. MHC-II−/− mice showed no bone defect by micro-CT (Fig. 7A) and no alteration of OCgenesis (Fig. 7B). The data indicate that the impact of CIITA on OC differentiation and bone homeostasis is independent of its well-known role in driving MHC-II expression.

Figure 7.

Normal bone microarchitecture and ex vivo osteoclastogenesis in MHC-II−/− mice. (A) Micro-CT analysis of proximal tibial epiphyses from 5-month-old MHC-II−/− and wild-type (WT) control littermates (n = 5 mice per group). Top panel: two representative 3D reconstructions. Bottom histograms: quantifications of trabecular bone volume/tissue volume (BV/TV), thickness (Tb.Th), number (Tb.N.), and separation (Tb.Sp.). Data expressed as mean ± SD. *p < 0.05. (B) OC differentiation assay on BMM purified from MHC-II−/− mice. Cells were differentiated in the presence of 10 ng/mL M-CSF and 100 ng/mL RANKL in 96-well plates, fixed, and stained for TRAP activity in triplicate wells at different time points. The left histogram shows the numbers of multinucleated TRAP+ cells (OC) after 7 days of culture. Data represent mean ± SD from three independent experiments, each pooling BMM from two littermates. Right panels show representative fields after 6 days of culture. Scale bar = 50 µm.

Enhanced osteoclastogenic signaling in CIITA overexpressing mice

To understand the molecular mechanisms by which CIITA positively regulates OCgenesis, we first asked if BMM from CIITA-overexpressing mutants differ in proliferation and apoptosis in basal conditions or during OC differentiation. We found no difference in proliferation, as assessed by CFSE dilution (Fig. 8A), and apoptosis, as indicated by nuclear morphology (Fig. 8B) and caspase 3 cleavage (Supplemental Fig. S4). We then analyzed the main signaling pathways that drive OC differentiation by immunoblotting against various phosphorylated and unmodified proteins representative of distinct OCgenic sub-pathways in CIITA-overexpressing BMM upon acute stimulation with M-CSF and RANKL. We found a number of signaling cascades downstream of c-fms and RANK to be altered in CIITA-overexpressing OC precursors. In particular, CIITA-overexpressing BMM showed slightly increased phosphorylation levels of ERK and more substantially increased phosphorylation of Akt upon stimulation with M-CSF (Fig. 8C, D). Moreover, despite normal surface expression of RANK (Supplemental Fig. S5), stimulation of mutant OC precursors with RANKL generated increased phosphorylation of JNK and ERK, and increased expression of the crucial OCgenic factor NFATc1 (Fig. 8E, F). Altogether, the observed increased responses to M-CSF and RANKL, associated with the enhanced OCgenic potential observed in vitro (Fig. 5) and in vivo (Fig. 3), suggest that CIITA positively regulates OC differentiation by stimulating OCgenic signaling, without affecting cell proliferation and death.

Figure 8.

Enhanced c-fms and RANK signaling, associated with normal proliferation and apoptosis, in CIITA-overexpressing precursors. (A) FACS analysis of CFSE 48-hour dilution in noncommitted BMM (top panel) and pre-OC, ie, BMM stimulated with 10 ng/mL M-CSF and 100 ng/mL RANKL for 3 days (bottom panel) from CIITA Tg mice and WT control littermates. One representative of three independent experiments is shown. (B) BMM from WT, CIITA Tg, and CIITA pIV−/− mice were induced to differentiate into OCs and nuclear morphology analyzed by confocal microscopy upon staining with Hoechst at the indicated time points. Pre-OC (day 3) and mature OC (day 10) were costained with Alexa Fluor 594 phalloidin to visualize OC perimeters and distinguish OC-like cells from mononucleated precursors. Scale bars = 25 µm. One representative of three independent experiments is shown. (C) Immunoblot analysis of c-fms signaling in WT and CIITA pIV−/− BMM upon stimulation with M-CSF (100 ng/mL) for the indicated times. One representative of three experiments is shown. (D) Relative densitometric quantification of the ratio of phosphorylated versus nonphosphorylated ERK and Akt 5 minutes post-M-CSF stimulation. (E) Immunoblot analysis of RANK signaling in pre-OC from WT and CIITA pIV−/− mice upon acute stimulation with 100 ng/mL RANKL for the indicated times. One representative of three experiments is shown. (F) Relative densitometric quantification of phosphorylated versus nonphosphorylated ratios of JNK and ERK (top histograms) and of NFATc1 (bottom graph, normalized versus β-actin) 60 minutes post-RANKL stimulation.

Discussion

A novel skeletal role of CIITA

Our in vivo findings unveil a novel, unexpected nonimmune function of CIITA—an established master transcriptional co-activator believed to be dedicated primarily for Ag presentation—as a potent regulator of bone homeostasis. This function appears to be mediated mainly through a stimulatory effect on OC differentiation and resorptive activity in the mature skeleton.

Whether CIITA-deficient BLS patients exhibit primary skeletal defects is unknown. The main clinical manifestations of this extremely rare and severe immunodeficiency (<100 patients from ∼60 unrelated families worldwide) are recurrent infections, malabsorption, failure to thrive, and death during childhood, usually around 4 years of age.[9, 35] The clinical severity of the disease, the rarity and short life expectancy of patients, and lack of evidence for nonimmune functions of CIITA have impeded the search for unanticipated manifestations of the disease, such as primary skeletal alterations.

The significance and implications of the in vivo skeletal role of CIITA described herein are manifold.

The bone homeostatic function of CIITA is T cell and MHC-II independent

T cells have been reported to exert important effects on bone physiology and disease,[1, 24, 36-38] raising the question as to whether they also play a role in the bone phenotype displayed by CIITA mutant mice. CIITA Tg and pIV−/− mice, the two models of CIITA overexpression adopted herein, respectively, systemic and restricted to the BMM lineage, show profound immune differences. Indeed, whereas CIITA Tg display a normal T cell repertoire,[30] mice lacking CIITA pIV, which is essential for MHC-II expression in thymic epithelial cells, lack CD4+ T cells.[19, 31] Thus, the observation that these two immunologically opposite mouse models have overlapping skeletal phenotypes (Figs. 2 and 3) strongly supports the T cell independence of the bone alterations observed. We conclusively demonstrated this point by analyzing an additional pIV−/− mouse line, pIV−/− K14 CIITA Tg mice, in which a CIITA transgene driven by the human keratin 14 promoter restores MHC-II expression in the thymic epithelial cortex and hence a normal CD4+ T cell population.[34] These mice revealed a trabecular defect similar to that observed in CIITA Tg and pIV−/− mice (Fig. 4), proving the T cell independence of the skeletal effects of CIITA overexpression. Moreover, the normal skeletal phenotype and OC differentiation displayed by MHC-II deficient mice (Fig. 7) indicate that the canonical transcriptional target of CIITA plays no skeletal role, and that CIITA's stimulatory effect on OC differentiation and bone resorption is independent of its established immune role in driving MHC-II expression.

CIITA and OCgenic signaling

The engagement of RANK by RANKL in presence of M-CSF activates a number of OCgenic pathways that lead to terminal OC differentiation.[39, 40] In search for the cellular and molecular bases whereby CIITA controls OCgenesis, we excluded an effect of CIITA overexpression on monocyte proliferation and programmed cell death, either under basal conditions or during OC differentiation, and identified a stimulatory effect of CIITA on different OCgenic sub-pathways downstream of c-fms and RANK signaling (Figs. 8 and S4). These data, together with evidence supporting a direct role of endogenous CIITA in OC precursors (Fig. 5) and opposing a role for BM SC in the enhanced OCgenesis observed in CIITA-overexpressing mice (Fig. S3), point to an intrinsic effect of CIITA on OCgenic signaling. Further studies are required to clarify if this action is mediated by CIITA or by a transcriptional target other than MHC-II. Interestingly, CIITA emerges as part of a novel molecular circuit linking the signaling pathways activated by M-CSF and RANKL.

By showing that CIITA is a positive regulator of RANK signaling, the data contradict a recent in vitro work by Kim and colleagues showing that overexpression of CIITA in BMM reduced OCgenesis.[41] The underlying mechanism proposed was sequestration of CBP/p300 by overexpressed CIITA at the expense of c-fos transactivating capacity, resulting in reduced induction of the crucial OCgenic factor, NFATc1. In light of our genetic evidence, the observed sequestration of general histone modifiers by an overexpressed transcriptional regulator in vitro may be the result of a general nonspecific effect on transcriptional programs.

RANKL also stimulates Ag presentation in DC by increasing co-stimulatory responses and pro-survival signals.[1-3, 42] In light of such immune function of RANKL, our novel observation that CIITA, the master switch of Ag presentation, stimulates RANK signaling (Fig. 8) raises the possibility of a role for a regulatory CIITA-RANK circuit also in DC. Although we found no effect of CIITA overexpression on DC differentiation, apart from the expected amplification of MHC-II expression (Fig. 6), by stimulating RANK signaling, CIITA may promote Ag presentation not only by driving MHC-II expression but also by prolonging DC life span and activity.

The skeletal function of CIITA is not isoform-specific

CIITA expression is regulated mainly at the level of transcription, although changes in mRNA and protein stability can contribute.[11] Transcription of CIITA is driven by distinct promoters (pI, pIII, and pIV), which precede alternative first exons that are spliced to shared downstream exons, giving rise to three different CIITA mRNAs and three protein isoforms (I, III, and IV), all able to restore MHC-II expression and to activate MHC-II genes.[11] While CIITA Tg mice overexpress type IV CIITA,[29, 30] CIITA pIV−/− mice overexpress through pI (Fig. 1). Thus, both CIITA type I and IV exert the novel pro-OCgenic and bone homeostatic effect reported herein.

Evolution and teleology of a skeletal function of CIITA

Our data may have implications for the common evolution of bone and immunity. The mineralized skeleton and adaptive immune system evolved relatively late in jawed vertebrates. The close anatomic proximity of bone and hematopoiesis suggests that the skeletal and immune systems may have co-opted differentiation and signaling strategies from each other.[43] Our discovery of a role for the master regulator of adaptive immunity CIITA in bone biology further supports this view. Moreover, phylogenetic evidence indicates that the appearance of an anticipatory immune response involving MHC-based allorecognition predates skeletal evolution, as witnessed by cartilaginous fishes, the most ancient jawed vertebrates,[44, 45] suggesting that bone may have co-opted CIITA as a regulator of OC differentiation from immunity.

Implications for bone diseases

Osteoporosis represents a major health problem worldwide, with a huge impact on morbidity and mortality, and a massive burden on the healthcare system.[46, 47] Despite significant progress, most genetic determinants of bone strength remain to be discovered.[48, 49] Because the CIITA locus is polymorphic, our work encourages the study of the association of its genetic variants with bone health, aiming to discover prognostic indicators and therapeutic targets. In keeping with our study, two CIITA variants were recently found to influence the rate of bone loss and fracture risk in elderly women.[50]

Our work also identifies CIITA-overexpressing mice as novel mouse models of uncoupled bone formation and resorption, resulting in spontaneous, early onset, severe osteoporosis, of obvious biomedical interest. In particular, because the observed osteopenia is selectively accounted for by increased OC numbers and activity (Fig. 3), these models are optimal for testing antiresorptive drugs in vivo. Moreover, as OB fail to compensate for increased bone resorption, these mice could also be valuable for evaluating the efficacy of osteoanabolic compounds.

In conclusion, despite the huge impact of osteoporosis worldwide, our knowledge of bone genetics and biology remains remarkably incomplete. Our findings and models provide compelling genetic evidence of a novel molecular network regulating adult skeletal homeostasis in vivo, which may help identify new specific therapeutic targets against involutional, inflammatory, or cancer-induced bone wasting.

Disclosures

All authors state that they have no conflicts of interest.

Acknowledgments

This study was supported by research grants to SC (European Calcified Tissue Society 2009 Career Establishment Award; Ministero della Salute, Giovani Ricercatori 2008-1143560 and 2009-1607545; Multiple Myeloma Research Foundation, 2010 Senior Research Award; and the Italian Association for Cancer Research, Special Program Molecular Clinical Oncology 5 per mille 9965). Work in WR's lab was supported by the Swiss National Science Foundation, the National Center of Competence in Research on Neural Plasticity and Repair (NCCR-NEURO), the Swiss Multiple Sclerosis Society, and the Geneva Cancer League. We are indebted to M Ascagni, M Baleani, C Covino, C Cucchi, A Del Fattore, S Dell'Oro, C Fersini, D Finke, F Grassi, N Lo Iacono, R Pacifici, MC Panzeri, N Pengo, P Podini, M Rocchi, P Rovere-Querini, N Rucci, C Sobacchi, S Takeshita, A Teti, A Villa, K Weilbaecher, and A Zallone for reagents and scientific advice. We are thankful to R Brambati, A Fella, and F Loro for technical and administrative assistance.

Authors' roles: EBe designed, performed, and analyzed most experiments; EMa performed histomorphometrical analyses; LO, MS, EMi, and UO performed selected biochemical, differentiation and imaging analyses; EP and FP performed micro–CT scans and analyses, with FB's and NF's supervision; LC and AC performed macrophage/DC differentiation experiments; EBa performed CIITA transcript analyses; LOt provided advice and generated mutant mice with HA–O; WR provided mouse models, technologies, supervision, and advice; RF and RS provided supervision and advice; SC conceived, designed, and supervised the study, analyzed results, and wrote the article.