CTLA-4Ig–Induced T Cell Anergy Promotes Wnt-10b Production and Bone Formation in a Mouse Model


  • The views expressed herein are those of the authors and do not represent the views of the Department of Veterans Affairs or the United States Government.

  • Emory University and the Atlanta VAMC have filed a patent application for the use of CD28 modulators for bone anabolic activity.



Rheumatoid arthritis (RA) is an inflammatory autoimmune disease characterized by severe joint erosion and systemic osteoporosis. Chronic T cell activation is a hallmark of RA, and agents that target the CD28 receptor on T cells, which is required for T cell activation, are being increasingly used as therapies for RA and other inflammatory diseases. Lymphocytes play complex roles in the regulation of the skeleton, and although activated T cells and B cells secrete cytokines that promote skeletal decline, under physiologic conditions lymphocytes also have key protective roles in the stabilization of skeletal mass. Consequently, disruption of T cell costimulation may have unforeseen consequences for physiologic bone turnover. This study was undertaken to investigate the impact of pharmacologic CD28 T cell costimulation blockade on physiologic bone turnover and structure.


C57BL6 mice were treated with CTLA-4Ig, a pharmacologic CD28 antagonist or with irrelevant control antibody (Ig), and serum biochemical markers of bone turnover were quantified by enzyme-linked immunosorbent assay. Bone mineral density and indices of bone structure were further measured by dual x-ray absorptiometry and micro–computed tomography, respectively, and static and dynamic indices of bone formation were quantified using bone histomorphometry.


Pharmacologic disruption of CD28 T cell costimulation in mice significantly increased bone mass and enhanced indices of bone structure, a consequence of enhanced bone formation, concurrent with enhanced secretion of the bone anabolic factor Wnt-10b by T cells.


Inhibition of CD28 costimulation by CTLA-4Ig promotes T cell Wnt-10b production and bone formation and may represent a novel anabolic strategy for increasing bone mass in osteoporotic conditions.

Rheumatoid arthritis (RA) is a chronic inflammatory autoimmune disease that leads to bone loss around inflamed joints, as well as a generalized systemic osteoporosis ([1-3]). Lymphocytes play central roles not only in the initiation and progression of the inflammatory state, but also in the bone loss associated with RA ([4-8]). Lymphocytes drive bone turnover as a consequence of the immuno–skeletal interface, an enigmatic centralization of immune and skeletal functions around common cell types and cytokine effectors ([9]). Immune cells, including T cells, B cells, and antigen-presenting cells (APCs), are implicated in the regulation of basal ([10]) and/or pathologic ([11]) bone turnover. Activated lymphocytes induce bone resorption by secreting RANKL, the key osteoclastogenic cytokine, and inflammatory factors including tumor necrosis factor α (TNFα), a key driver of inflammatory cascades in RA. In addition, activated T cells produce secreted osteoclastogenic factor of activated T cells, a RANKL-independent osteoclastogenic cytokine that may contribute to bone loss in RA ([12, 13]) and in periodontal infection ([14]).

In contrast, under physiologic conditions lymphocytes are protective of the skeleton, as both human ([15]) and rodent ([9, 10]) B cells secrete the RANKL decoy receptor osteoprotegerin (OPG). Because T cell costimulatory interactions amplify B cell OPG production ([10, 15]) disruptions to adaptive immune function can lead to RANKL/OPG imbalances that are permissive for osteoclastogenesis. Indeed, alterations to the immuno–skeletal interface causing a B cell inversion in OPG and RANKL production may account, in part, for the bone loss characteristic of human immunodeficiency virus infection ([9, 16, 17]).

T cells express several unique receptors/ligands necessary for immune regulation including the CD28 receptor, which binds to CD80/CD86 ligands expressed by APCs and mediates signals necessary for T cell activation following binding of the T cell receptor (TCR) to the antigen-bearing major histocompatibility complex (MHC). Failure to activate CD28, or inhibition of CD28 signaling by CTLA-4, a physiologic modulator that is homologous with CD28 and competes for its ligands, leads to abortive T cell activation and/or terminates immune responses, resulting in T cell anergy or deletion ([18, 19]).

CTLA-4Ig (abatacept), an antiinflammatory pharmaceutical agent comprising the binding domain of human CTLA-4 fused to human IgG1, is approved for the treatment of refractory RA in adults ([20]) and juvenile idiopathic arthritis in children ([21]). Our group has reported that CTLA-4Ig mitigates ovariectomy-induced bone loss by reducing T cell activation and expression of TNFα via disruption of communication between T cells and dendritic cells ([22]). Similarly, CTLA-4Ig ameliorates bone loss in mice treated with continuous infusion of parathyroid hormone (PTH), a model of hyperparathyroidism ([23]). Furthermore, CTLA-4Ig has been reported to directly suppress osteoclast differentiation in the absence of T cells in vitro and to inhibit inflammatory bone erosion in vivo in an animal model of RA ([24]).

Because CTLA-4Ig disrupts costimulatory interactions between B cells and T cells, it has the potential not only to reduce immune activation responsible for driving inflammation, but also to disrupt basal bone turnover by disturbing the immune–skeletal interface and B cell OPG production. This effect could potentially offset the gains in bone mass associated with reduced inflammation. In this study we investigated the net effect of CTLA-4Ig on basal bone turnover and mass in mice, by quantifying indices of bone structure and turnover. CTLA-4Ig treatment led to significant bone accrual but surprisingly, this was a result of increased bone formation, as a likely consequence of T cell expression of the bone anabolic ligand Wnt-10b. Our data show for the first time that CTLA-4Ig leads to induction of bone formation and may have potential applications as a novel bone anabolic agent.



All reagents were purchased from Sigma-Aldrich Chemical unless otherwise indicated.


All animal studies were approved by both the Atlanta Veterans Administration Medical Center and Emory University Animal Care and Use Committees and were conducted in accordance with the National Institutes of Health Laboratory Guide for the Care and Use of Laboratory Animals. Mice were housed under specific pathogen–free conditions and were fed gamma-irradiated 5V02 mouse chow (Purina Mills) and autoclaved water ad libitum. The animal facility was kept at 23°C (±1°C) with 50% relative humidity and a 12 hour/12 hour light/dark cycle.

Young (6-week-old) female C57BL6 wild-type and ovalbumin-specific class I MHC–restricted T cell (OTI) mice were from The Jackson Laboratory, and skeletally mature (5-month-old) mice were from the National Institute on Aging aged mouse colony at Charles River Laboratories. Wild-type mice were injected intraperitoneally with 10 mg/kg CTLA-4Ig (Orencia; Bristol-Myers Squibb) or human Ig (Lampire Laboratories) twice weekly for 12 weeks or 26 weeks. Of the skeletally mature mice treated for 12 weeks, 8 received the Ig treatment and 7 received the CTLA-4Ig treatment. Skeletally mature mice treated for 26 weeks also comprised 8 in the Ig treatment group and 7 in the CTLA-4Ig group. Of the young mice treated for 12 weeks, there were 10 animals per treatment group. Young mice treated for 6 months comprised 12 mice per treatment group; however, 1 extreme outlier in the CTLA-4Ig group, with a final mean bone volume/total volume (BV/TV) of 0.12 (3 SD below the mean, and well below even the mean value of 2.07 in the wild-type mice), was eliminated from the micro–computed tomography (micro-CT) and histomorphometric analyses. One bone used for histomorphometry was damaged during processing and had no quantifiable bone or cells.

Bone densitometry

Bone mineral density (BMD) (gm/cm2) was quantified in anesthetized mice by dual x-ray absorptiometry (DXA) using a Piximus II bone densitometer (GE Medical Systems). Total-body DXA was performed and region-of-interest boxes placed to quantify findings at specific anatomic sites, including the lumbar spine, femur, and tibia, as previously described ([25]). Measurements from the left and right femora and left and right tibiae were averaged for each mouse, and the mean used for group calculations.


Micro-CT was performed in L3 vertebrae and femoral metaphysis ex vivo to assess trabecular bone microarchitecture using a μCT40 scanner (Scanco Medical) that was calibrated weekly with a factory-supplied phantom. A total of 405 tomographic slices were taken at the L3 vertebra (total area of 2.4 mm) and 100 tomographic slices at the distal femoral metaphysis and trabecular bone segmented from the cortical shell, for a total area of 0.6 mm beginning ∼0.5 mm from the distal growth plate. Projection images were reconstructed using the autocontour function for trabecular bone. Cortical bone was quantified at the mid–femoral diaphysis from 100 tomographic slices. Representative vertebral samples based on mean BV/TV were reconstructed 3-dimensionally (3-D) to generate visual representations. Indices and units were standardized according to published guidelines ([26]).

Quantitative bone histomorphometry

Bone histomorphometric analysis was performed at the University of Alabama at Birmingham Center for Metabolic Bone Disease–Histomorphometry and Molecular Analysis Core Laboratory. Trichrome-stained plastic-embedded sections of calcein-labeled femora from Ig- and CTLA-4Ig–injected mice were used for these analyses.

Enzyme-linked immunosorbent assays (ELISAs) for biochemical indices of bone turnover and for Wnt-10b

C-terminal telopeptide of collagen (CTX) and osteocalcin in mouse serum were quantified using RATlaps and Rat-MID ELISAs, respectively (both from Immunodiagnostic Systems). Levels of Wnt-10b protein in 24-hour conditioned media from negatively immunomagnetically purified CD3 T cells (Miltenyi Biotech) were determined using a Wnt-10b ELISA (USCN Life Science).

Real-time reverse transcription–polymerase chain reaction (RT-PCR).

Total RNA was extracted from whole nucleated bone marrow, flushed from long bones, and dissolved in TRIzol reagent. Real-time RT-PCR was performed on an ABI Prism 7000 (Applied Biosystems) as previously described ([10]), using commercial (Applied Biosystems) Master Mix and primer sets and probes for murine Wnt-10b (catalog no. Mm 00442104), OPG (catalog no. Mm 001205928), RANKL (catalog no. Mm 00441906), and β-actin (catalog no. Mm 00607939). Changes were calculated using the 2−ΔΔCt method ([27]) with normalization to β-actin.

T cell activation assays

Plates were coated overnight with activating anti-mouse CD3e antibodies (5 μg/ml) and/or anti-mouse CD28 antibodies (25 μg/ml) in sterile phosphate buffered saline (eBioscience). Splenic T cells were isolated using a CD3 Pan T cell isolation kit (Miltenyi Biotec). Twelve replicate wells were plated at 2 × 107cells/well in 24-well plates in 750 μl RPMI 1640/5% fetal bovine serum (FBS) for 24 hours, and then dissolved in TRIzol for RNA isolation and real time RT-PCR for assessment of Wnt-10b expression as described above.

APC assays

APC assays were performed as previously described ([28]), with modifications. Briefly, immunomagnetically purified (Miltenyi Biotech) splenic CD11c dendritic cells were used as APCs. Cells were plated in triplicate at 150,000/well in complete RPMI 1640/10% FBS and incubated for 4 hours at 37°C with 1 μM antigen (ovalbumin [OVA] peptide) (SIINFEKL; InvivoGen), followed by 2 washes in medium. CD8+ T cells expressing a monoclonal OVA-specific transgenic TCR were purified from the spleens of OTI mice, and 1 million T cells were incubated for 24 hours with OVA-presenting APCs with or without CTLA-4Ig. T cells and dendritic cells were dissolved in TRIzol for RNA isolation and real time RT-PCR for determination of Wnt-10b expression.

Statistical analysis

Analyses were performed using GraphPad InStat version 3.0 for Windows (GraphPad Software). Gaussian distribution was assessed by Kolmogorov-Smirnov test. The significance of differences was assessed using Student's 2-tailed t-test or the Mann-Whitney test for nonparametric data. Group comparisons were made by one-way analysis of variance with Tukey-Kramer post hoc test. P values less than or equal to 0.05 were considered significant.


Elevated BMD and enhanced trabecular and cortical bone volume in CTLA-4Ig–treated mice

To investigate the net effect of the immunosuppressive agent CTLA-4Ig on physiologic basal bone modeling, we injected CTLA-4Ig or irrelevant isotype control Ig into young (6-week-old) female C57BL6 mice. BMD was examined after 12 weeks and 26 weeks by bone densitometry using DXA. Although CTLA-4Ig administration failed to cause any significant change in BMD compared to Ig control treatment at 3 months (Figure 1A), by 6 months of treatment CTLA-4Ig–injected mice displayed a significant increase in total-body BMD and increases at specific anatomic sites including the femur and tibia (left and right femur or tibia for each mouse independently averaged) and lumbar spine (Figure 1B).

Figure 1.

Increased bone mineral density (BMD) and bone mass in young and skeletally mature mice treated with CTLA-4Ig. Total-body BMD and BMD at the lumbar spine, femur, and tibia were quantified by dual x-ray absorptiometry in young mice 3 months (A) and 6 months (B) after administration of Ig (control) (open bars) or CTLA-4Ig (solid bars) and in skeletally mature mice 6 months after administration of Ig or CTLA-4Ig (C). Values are the mean ± SEM (n = 10 Ig-treated and 10 CTLA-4Ig–treated mice in A, 12 Ig-treated and 11 CTLA-4Ig–treated mice in B, and 8 Ig-treated and 7 CTLA-4Ig–treated mice in C). Δ = difference from the value in Ig-treated mice. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001, versus Ig-treated mice. NS = not significant. Representative high-resolution (6-μm) micro–computed tomography (μCT) reconstructions of vertebral cancellous bone in Ig- or CTLA-4Ig–treated mice from each group are shown below the graphs. Bars = 500 μm.

To assess the effect of long-term CTLA-4Ig treatment on the remodeling skeletons of adult mice, we further treated 5-month-old mice with CTLA-4Ig or control Ig for 6 months. As with young mice, total-body BMD in 5-month-old mice treated with CTLA-4Ig was significantly elevated, as was BMD at the femur, lumbar spine, and tibia (Figure 1C).

To independently evaluate cortical and cancellous bone, we examined the lumbar vertebrae and femora by micro-CT. Representative micro-CT reconstructions of vertebral trabecular bone in young skeletally immature mice treated with CTLA-4Ig for 3 months, young mice treated for 6 months, and skeletally mature mice treated for 6 months are shown in Figures 1A–C.

Quantitative microarchitectural indices of trabecular bone structure in young mice receiving 3-month or 6-month CTLA-4Ig treatment (Table 1) and in mature mice receiving 6-month treatment (Table 2) were determined. Vertebral trabecular BV/TV was significantly increased in young mice receiving CTLA-4Ig for either 3 months or 6 months, consistent with a decline in trabecular separation (TbSp), reflecting the amount of bone-free space. Trabecular thickness (TbTh) and trabecular number (TbN) were significantly increased in young mice by 6 months of CTLA-4Ig treatment but the difference from controls fell just short of significance at 3 months, suggesting a slow accumulation of bone volume over time. As expected, bone accretion in mature mice was slower than in younger animals, and the difference in BV/TV between CTLA-4IG–treated and control mice was not quite statistically significant, although TbN and TbSp values differed significantly between the 2 groups.

Table 1. Micro-CT–assessed bone structure indices in the L3 vertebrae and femora of young mice administered Ig (control) or CTLA-4Ig for 3 months or 6 months*
Treatment duration, site, indexIgCTLA-4Ig% changeP
  1. Values are the mean ± SD (n = 10, 3-month Ig treatment and 3-month CTLA-4Ig treatment; n = 12, 6-month Ig treatment; n = 11, 6-month CTLA-4Ig treatment). Micro-CT = micro–computed tomography; BV/TV = bone volume/total volume; TbTh = trabecular thickness; TbN = trabecular number; TbSp = trabecular separation; CoAr = cortical area; CoTh = cortical thickness.
3 months    
L3 vertebra    
BV/TV, %13.15 ± 1.2515.07 ± 0.9514.60.0011
TbTh, mm0.0390 ± 0.00100.0406 ± 0.00244.00.0761
TbN, mm3.71 ± 0.203.91 ±
TbSp, mm0.2675 ± 0.01470.2526 ± 0.0161−5.60.0439
BV/TV, %6.35 ± 1.358.04 ± 1.2218.80.0088
TbTh, mm0.0412 ± 0.00260.0433 ± 0.00374.90.1490
TbN,/mm3.44 ± 0.343.69 ± 0.32−0.40.1091
TbSp, mm0.2822 ± 0.01470.2713 ± 0.02472.00.2451
CoAr, mm20.8074 ± 0.03220.8068 ± 0.0339−0.10.9696
CoTh, mm0.1840 ± 0.00540.1864 ± 0.00651.30.3830
6 months    
L3 vertebra    
BV/TV, %13.33 ± 1.0516.91 ± 1.7626.9<0.0001
TbTh, mm0.0450 ± 0.00270.0474 ± 0.00265.30.0447
TbN,/mm3.15 ± 0.123.53 ± 0.3112.1<0.0001
TbSp, mm0.3195 ± 0.01550.2843 ± 0.0222−11.00.0002
BV/TV, %3.24 ± 1.014.42 ± 1.3236.60.0262
TbTh, mm0.0442 ± 0.00450.0485 ± 0.00609.70.0631
TbN,/mm2.67 ± 0.192.76 ±
TbSp, mm0.3791 ± 0.02730.3666 ± 0.0243−3.30.2640
CoAr, mm20.7462 ± 0.02680.7835 ± 0.05975.00.0630
CoTh, mm0.1898 ± 0.00720.2000 ± 0.01045.40.0114
Table 2. Micro-CT–assessed bone structure indices in the L3 vertebrae and femora of skeletally mature mice administered Ig (control) or CTLA-4Ig for 6 months*
Site, indexIg (n = 8)CTLA-4Ig (n = 7)% changeP
  1. Values are the mean ± SD. See Table 1 for definitions.
L3 vertebra    
BV/TV, %11.74 ± 1.2613.31 ± 1.8513.40.0730
TbTh, mm0.0508 ± 0.00340.0492 ± 0.0045−3.20.4324
TbN,/mm2.72 ± 0.122.94 ±
TbSp, mm0.3658 ± 0.01700.3409 ± 0.0222−6.80.0292
BV/TV, %0.82 ± 0.481.11 ± 0.6736.10.3275
TbTh, mm0.0506 ± 0.01110.0406 ± 0.0128−19.70.1181
TbN,/mm2.35 ± 0.142.39 ±
TbSp, mm0.4197 ± 0.10750.4463 ± 0.03536.30.5171
CoAr, mm20.7716 ± 0.03140.7940 ± 0.04192.90.2464
CoTh, mm0.1774 ± 0.00850.1769 ± 0.0135−0.30.9306

Femoral trabecular BV/TV was significantly increased after both 3 months and 6 months of CTLA-4Ig treatment in young animals, although changes in TbSp, TbTh, and TbN did not achieve statistical significance. At both time points TbTh showed the largest increase, just narrowly falling short of significance at 6 months of treatment. These data suggest that CTLA-4Ig may work predominantly by expanding the thickness of preexisting trabecular spicules rather than catalyzing de novo synthesis of new template. In accordance with the age of the mature mice at the end of the treatment period (11 months), there was very little trabecular bone remaining in the femora. Despite inconsistency in results for the structural indices, a large mean increase in BV/TV was observed with CTLA-4Ig treatment. However, due to high variability within the groups, no statistically significant changes were achieved.

Cortical bone was quantified in the mid–femoral diaphysis, revealing a significant increase in cortical thickness in young mice treated with CTLA-4Ig for 6 months (Table 1). Cortical area was also increased by 6 months, but the increase was not quite statistically significant. Neither index changed significantly in mature mice after treatment. Because total-body BMD and BMD at all individual anatomic sites assessed by DXA increased significantly after treatment in mature animals (Figure 1A), the data suggest that small increases in cortical bone across large areas may account for the increases demonstrated by DXA but not by micro-CT.

Enhancement of biochemical indices of bone formation, but not indices of bone resorption, in vivo in CTLA-4Ig–treated mice

To assess rates of bone resorption and bone formation in vivo, bone turnover markers in the serum of CTLA-4Ig–treated and Ig-treated mice were quantified. Levels of CTX, a sensitive and specific index of bone resorption, were not significantly different between CTLA-4Ig–treated and control mice of either young age group, irrespective of the length of treatment (among young mice, mean ± SD 24.7 ± 5.7 ng/ml and 23.8 ± 9.4 ng/ml after 6 months of control Ig treatment and 6 months of CTLA-4Ig treatment, respectively; among mature mice, 13.2 ± 5.9 ng/ml and 14.2 ± 3.3 ng/ml, respectively).

Consistent with these data, RT-PCR analysis of bone marrow revealed no significant alterations in expression of OPG or RANKL after 6 months of treatment in young mice (OPG 2−ΔΔCt 1.00 ± 0.18 and 0.87 ± 0.17 in control Ig–treated mice and CTLA-4Ig–treated mice, respectively; RANKL 2−ΔΔCt 1.00 ± 0.04 and 0.88 ± 0.05, respectively). In contrast, serum osteocalcin, a biochemical marker of in vivo bone formation, was dramatically increased in young mice treated with CTLA-4Ig for 3 months or 6 months and in mature mice treated for 6 months (Figure 2A). These results suggest that bone accretion was a consequence of elevated bone formation.

Figure 2.

Promotion of bone formation in mice by CTLA-4Ig treatment. A, Levels of osteocalcin, a serum marker of bone formation, in young and skeletally mature mice treated with Ig (control) or CTLA-4Ig for 3 months or 6 months. Values are the mean ± SD. ∗∗∗ = P < 0.001 versus Ig-treated mice. B, Representative images showing calcein double-fluorescence labeling, used to compute mineral apposition and bone formation rates, in femora from young mice treated with Ig or CTLA-4Ig for 6 months. Original magnification × 40. C, Representative histologic sections of Goldner's trichrome–stained distal femora from young mice treated with Ig or CTLA-4Ig for 6 months. Mineralized bone stains green. Trabecular bone in bone marrow cavity is indicated by yellow arrows, and epiphyseal bone by red arrows. Original magnification × 100.

Enhancement of histomorphometric indices of bone formation in CTLA-4Ig–treated mice

As osteocalcin reflects global bone formation across all bone surfaces, we assessed bone formation at the tissue and cellular level in young mice treated for 6 months with CTLA-4Ig, by quantitative bone histomorphometry. Three indices of bone formation, i.e., mineralized surface (MS), mineral apposition rate (MAR), and bone formation rate (BFR)/TV, were all significantly increased with treatment (Table 3). When MS and BFR were normalized to bone surface, BFR was increased to a level just short of significance, while BFR and MS normalized to BV were unchanged from control.

Table 3. Histomorphometric findings in the femora of young mice administered Ig (control) or CTLA-4Ig for 6 months*
 Ig (n = 12)CTLA-4Ig (n = 10)% changeP
  1. Values are the mean ± SD. BS = bone surface; MS = mineralizing surface; MAR = mineral apposition rate; BFR = bone formation rate; ObS = osteoblast surface; NOb = number of osteoblasts; OcS = osteoclast surface; NOc = number of osteoclasts (see Table 1 for other definitions).
  2. aCalculated as [(dLS + [sLS/2])/BS] × 100, where dLS = double-labeled surface and sLS = single-labeled surface.
Structure indices    
BV/TV, %2.07 ± 0.833.96 ± 1.2791.40.0004
BS, mm2.40 ± 1.194.13 ± 1.1471.90.0025
TbTh, mm0.0277 ± 0.00370.0305 ± 0.004010.20.1039
TbN,/mm0.76 ± 0.321.28 ± 0.3168.90.0010
TbSp, mm1.679 ± 1.1670.812 ± 0.279−51.60.0333
Dynamic bone formation indices    
MS, mm1.42 ± 0.582.01 ± 0.73641.50.0148
MS/BS, %a42.59 ± 8.1844.11 ±
MAR, μm/day1.44 ± 0.251.62 ± 0.1913.00.0443
BFR/BS, μm/day0.607 ± 0.1360.717 ± 0.15118.10.0870
BFR/TV,/day0.0013 ± 0.00070.0020 ± 0.000751.00.0315
BFR/BV,/day0.0394 ± 0.00870.0394 ± 0.00190.20.9879
Osteoblast indices    
ObS/BS26.98 ± 15.8221.49 ± 9.74−20.40.3555
NOb/BS14.02 ± 7.9512.19 ± 5.60−13.00.5478
Osteoclast indices    
OcS/BS6.81 ± 3.125.35 ± 2.49−21.50.2453
NOc/BS1.98 ± 0.781.89 ± 0.92−4.60.8051

The static indices osteoblast surface/bone surface and number of osteoblasts/bone surface showed a nonsignificant decline after treatment, indicating that the long-term effect of CTLA-4Ig was not to increase osteoblast number but rather to induce activation of preexisting osteoblasts, driving a wave of new bone formation. Consequently, the number of osteoblasts and the area of bone covered by osteoblasts both appeared to decline as a result of the significantly increased bone surfaces. BFR/TV, which normalizes for total bone area, is the index that correlates most closely with bone turnover markers such as osteocalcin ([29]). Similarly, small nonsignificant declines in osteoclast number/bone surface and osteoclast surface/bone surface were observed, likely also reflecting the relative increase in bone surface, rather than a long-term direct effect of CTLA-4Ig on osteoclast numbers.

Cancellous structural indices, including BV, BV/TV, TbN, and TbSp as determined by histomorphometry, all showed robust changes, supporting the notion of an increase in bone mass with CTLA-4Ig treatment as observed in the micro-CT studies. Surprisingly, the percentage change in BV/TV of femoral bone as determined by histomorphometry was more than twice that observed by micro-CT. In principle, micro-CT is a more robust and reliable measure of bone volume and structure than histomorphometry since micro-CT reflects a true 3-D quantification of a relatively large segment of bone while histomorphometry is a 2-D representation of a comparatively small number of slices. This diminished precision likely accounts for the apparent discrepancy.

Calcein double-fluorescence labeling, from which MAR and BFR indices were calculated, revealed enhanced bone formation in young CTLA-4Ig–treated mice after 6 months of treatment (Figure 2B). Goldner trichrome–stained femoral sections exhibited enhanced numbers of trabecular elements in the femoral metaphysis and increased bone thickness in the epiphyses above the growth plate (Figure 2C).

Elevated levels of the anabolic Wnt ligand Wnt-10b in total bone marrow and purified T cells from CTLA-4Ig–treated mice

T cells have the capacity to secrete Wnt-10b, a potent bone anabolic Wnt ligand ([30]). To investigate a possible role of Wnt-10b in the anabolic activity of CTLA-4Ig, we quantified Wnt-10b expression in the bone marrow of CTLA-4Ig– and Ig-treated mice by real time RT-PCR, and in conditioned medium from purified CD3+ T cells by ELISA (Figure 3A). Expression of Wnt-10b in the bone marrow of CTLA-4Ig–treated animals, as well its protein production by purified CD3+ T cells from treated animals, were elevated, suggesting involvement of Wnt-10b produced from T cells in the anabolic activity of CTLA-4Ig.

Figure 3.

The bone anabolic Wnt ligand Wnt-10b is potently up-regulated by CTLA-4Ig in vivo and suppressed by CD28 activation in vitro. A, Wnt-10b production in Ig (control)– and CTLA-4Ig–treated mice, quantified in whole bone marrow by real time reverse transcription–polymerase chain reaction (RT-PCR) and in conditioned media from purified T cells by enzyme-linked immunosorbent assay. Values are the mean ± SEM (n = 4 mice per group). ∗ = P < 0.05 versus Ig-treated mice. B, Wnt-10b expression in purified T cells activated by CD3 antibody with and without CD28-activating antibody, quantified by RT-PCR. Values are the mean ± SD of 12 individual wells per group and are representative of 2 independent experiments. ∗∗∗ = P < 0.001. C, Wnt-10b expression in T cells activated in vitro by antigen-presenting cells (APCs) with or without CTLA-4Ig, quantified by RT-PCR. Values are the mean ± SD of 3 individual wells per group and are representative of 2 independent experiments. ∗∗∗ = P < 0.001. D, Proposed model of the mechanism for the anabolic response of CTLA-4Ig, involving Wnt-10b expression by T cells. MHC = major histocompatibility complex; PDE = phosphodiesterase; TCR = T cell receptor; LRP5/6 = low-density lipoprotein receptor–related protein 5/6.

CD28 inhibits Wnt-10b expression by activated T cells in vitro, while CTLA-4Ig suppression of CD28 signaling amplifies Wnt-10b expression induced by antigen presentation in vitro

To further explore the mechanism of T cell Wnt-10b production induced by CTLA-4Ig, we purified T cells and activated them in vitro using CD3e-activating antibody in the presence or absence of CD28-activating antibody. Activation of CD3 led to a significant up-regulation of Wnt-10b expression at 24 hours (Figure 3B). Activation of CD28 alone had no significant effect on Wnt-10b, but potently inhibited Wnt-10b expression induced by CD3. CTLA-4Ig may thus promote Wnt-10b expression in T cells by blocking the interaction of CD80/CD86 on APCs with T cell–expressed CD28, a negative signal for Wnt-10b expression. To test this hypothesis directly, we performed an in vitro APC assay in which purified dendritic cells were used as APCs to express OVA antigen to CD8+ T cells expressing a monoclonal TCR with OVA-specific recognition. Presentation of OVA to T cells by APCs led to induction of Wnt-10b expression, which was potently superinduced by addition of CTLA-4Ig to the culture (Figure 3C).


Our data demonstrate for the first time that pharmacologic suppression of CD28 costimulation by CTLA-4Ig results in a bone anabolic signal, which is likely a consequence of Wnt-10b secretion by T cells. Although the capacity of the immune system to regulate bone resorption though perturbations of the immuno–skeletal interface is well studied, little is known regarding the immune system's ability to regulate bone formation. Activated T cells have been previously reported to secrete cocktails of cytokines that cumulatively have the capacity to induce alkaline phosphatase activity in purified human bone marrow stromal cells and elevate expression of Runx2 and osteocalcin messenger RNA ([31]). In contrast, a number of cytokines involved in immune regulation, such as interleukin-7 ([32]), or produced by immune cells including T cells and macrophages, such as TNFα ([33]), may act to uncouple bone formation from resorption under inflammatory conditions. Thus, with suppression of the natural compensatory increase in bone formation in response to increased bone resorption, bone homeostasis is disrupted, leading to bone loss ([11, 32]). Indeed, CTLA-4Ig has been reported to ameliorate short-term ovariectomy-induced bone loss by reducing T cell activation and inflammatory cascades ([22]) and to directly suppress osteoclast differentiation in vitro and inflammatory bone erosion in vivo in an animal model of TNFα-induced arthritis ([24]).

The molecular mechanism by which CTLA-4Ig stimulates Wnt-10b production remains to be studied in detail. However, based on our data and previously published reports, we propose a model whereby Wnt-10b production is an unintended consequence of abortive T cell activation due to disruption of the dual-signal mechanism of T cell activation. It is now established that two signals are required for full T cell activation. The first signal is generated when the TCR engages class I or II MHC–bearing antigens on the surface of professional APCs (B cells, macrophages, and dendritic cells). This first signal is insufficient for T cell activation, and on its own simply renders T cells unresponsive to further antigenic stimuli (anergy) ([18]). At the molecular level this signal involves activation of the cAMP second messenger system. The generation of cAMP leads to activation of protein kinase A and CREB, a transcription factor that transactivates genes involved in T cell regulation and anergy. This costimulatory signal involves binding of the CD28 receptor on T cells with B7 molecules (B7-1 [CD80] and B7-2 [CD86]) expressed on professional APCs (including B cells, macrophages, and dendritic cells). Activation of CD28 leads to induction of phosphodiesterase, an enzyme that cleaves cAMP, neutralizing its second messenger activity and eliminating the repressive first signal, thus preventing anergy and allowing full T cell activation to proceed ([34]). These two signals lead to full T cell activation, cytokine production, clonal expansion, and prevention of anergy ([18]).

In the context of normal bona fide immunologic actions cAMP generation would be shortlived, as the “verification signal” transduced though CD28 would rapidly shut off this pathway, preventing release of Wnt-10b. However, in the context of an impeded CD28 signal T cell cAMP production would remain active, leading to Wnt-10b secretion and binding to Wnt receptors (low-density lipoprotein receptor–related protein 5 [LRP-5] and LRP-6) on osteoblasts—in turn resulting in their activation and new bone formation. This model is depicted schematically in Figure 3D and is further supported by published gene array data demonstrating that Wnt-10b expression is up-regulated in anergic T cells ([35]).

Interestingly, we have previously demonstrated that the anabolic activity of intermittently administered PTH is mediated in part though T cell production of Wnt-10b ([30, 36]), a Wnt ligand that has been reported to be secreted by T cells ([37, 38]). Because PTH is a potent inducer of cAMP, we hypothesize that Wnt-10b expression may be directly promoted by activation of cAMP in T cells, bypassing normal TCR-mediated interactions with APCs (TCR and CD28 signaling) and leading to potent sustained production of Wnt-10b. An interesting conundrum that would be explained by this hypothesis is the question of why genetic deletion of T cells leads to an increase in bone resorption ([10, 39, 40]) but fails to dramatically curtail bone formation ([30]), while CD28 inhibition alone induces bone formation. We speculate that Wnt-10b production is elicited only under conditions involving a significant impediment to CD28 signaling (such as exogenous application of CTLA-4Ig or intermittent administration of PTH). Consequently, under physiologic conditions, basal T cell activation is relatively weak, and hence little Wnt-10b is secreted. Even under inflammatory conditions, CD28 activation during true APC-mediated T cell activation would quickly silence Wnt-10b expression.

Our group has reported that CTLA-4Ig protects against bone loss induced by ovariectomy ([22]) or continuous PTH administration ([23]) by blunting osteoclastic bone resorption driven by inflammatory cytokines, such as RANKL and TNF, secreted by T cells. Anabolic effects were not observed in these relatively short-term studies, and in fact our current data suggest that the anabolic effect of CTLA-4Ig is gradual but progressive, with significant bone gains thus achieved over an extended period of time. Although early changes in resorption may also occur following CTLA-4Ig treatment, the present results suggest that the major net effect of long-term administration, as used therapeutically in humans, is likely to be predominantly anabolic.

Consistent with the high basal bone turnover in young mice, bone formation and bone accretion were robust in young mice treated with CTLA-4Ig. In contrast, bone accretion in skeletally mature mice was more modest, and whereas we observed a strong trend toward bone gain, the change in BV/TV in the vertebrae was just short of statistically significant, although some structural indices, including TbN and TbSp, were significantly changed. Significant increases were not observed in the femur, likely due to the rapid degradation of trabecular bone in the femora of mice following peak BMD, leaving a denuded template for osteoblasts to act on. Interestingly, although basal bone formation represented by serum osteocalcin was significantly diminished in skeletally mature mice compared to younger mice, CTLA-4Ig did potently promote bone formation in these mice, suggesting that an anabolic response was in fact under way and that a statistically significant response would be likely over a longer period of time.

The capacity of CTLA-4Ig to promote bone formation in humans remains to be demonstrated; however, the use of this agent to treat inflammatory diseases such as RA may have multiple beneficial effects, including reduced inflammation and reduced osteoclastic bone resorption driven by inflammation ([22]) as well as due to direct inhibitory effects on osteoclasts ([24]). The present findings further suggest that bone anabolic activities due to T cell release of Wnt-10b may further promote a beneficial balance between bone formation and resorption.

In contrast to findings in inflammatory conditions, our studies of basal bone turnover did not reveal significant declines in CTX, an index of bone resorption, following long-term treatment with CTLA-4Ig. This was unexpected given that CTLA-4Ig has been reported to exert a direct antiosteoclastic effect in vitro ([24]). A possible explanation is that acute effects on osteoclasts are indeed observed early in the treatment, but with a return to baseline over time.

In conclusion, we have demonstrated that CTLA-4Ig, a CD28 costimulation inhibitor, is a potent bone anabolic agent and promotes the production of Wnt-10b by T cells. While there are numerous antiresorptive drugs currently available for the treatment of osteoporotic conditions, there are very few anabolic agents. Currently teriparatide, a fragment of PTH administered intermittently, is the only Food and Drug Administration–approved modality for stimulating bone accretion. Abatacept may represent a novel anabolic agent that could potentially be repurposed to ameliorate osteoporosis by stimulating bone formation, either as a monotherapy or in combination with other anabolic or anticatabolic agents.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Weitzmann had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Weitzmann.

Acquisition of data. Roser-Page, Vikulina, Zayzafoon.

Analysis and interpretation of data. Roser-Page, Vikulina, Zayzafoon, Weitzmann.