Article Text

This article has a correction. Please see:

PDF

CD4+CD45RBHi T cell transfer induced colitis in mice is accompanied by osteopenia which is treatable with recombinant human osteoprotegerin
  1. F R Byrne1,
  2. S Morony2,
  3. K Warmington2,
  4. Z Geng2,
  5. H L Brown1,
  6. S A Flores1,
  7. M Fiorino1,
  8. S L Yin1,
  9. D Hill2,
  10. V Porkess2,
  11. D Duryea3,
  12. J K Pretorius3,
  13. S Adamu3,
  14. R Manuokian4,
  15. D M Danilenko3,
  16. I Sarosi3,
  17. D L Lacey3,
  18. P J Kostenuik2,
  19. G Senaldi1
  1. 1Department of Inflammation, Amgen Inc., Thousand Oaks, California, USA
  2. 2Department of Metabolic Disorders, Amgen Inc., Thousand Oaks, California, USA
  3. 3Department of Pathology, Amgen Inc., Thousand Oaks, California, USA
  4. 4Department of Pre-clinical and Protein Therapeutics, Amgen Inc., Thousand Oaks, California, USA
  1. Correspondence to:
    Dr F Byrne
    Amgen Inc, One Amgen Center Drive, Mail Stop 29-1-B, Thousand Oaks, California 91320, USA; fbyrneamgen.com

Abstract

Background and aims: Transfer of CD4+CD45RBHi T cells into semi syngeneic immunodeficient mice represents a model of inflammatory bowel disease (IBD). As patients with IBD often suffer from osteopenia, we studied if this T cell transfer in mice results in osteopenia in addition to colitis, and if treatment with osteoprotegerin (OPG) has effects on the bone mineral density of T cell transferred mice. We also investigated whether osteopenia was due to malabsorption as a result of a dysregulated digestive tract or as a consequence of the inflammatory process.

Methods: CD4+CD45RBHi or CD4+CD45RBLo T cells (4×105) were sorted from CB6F1 and transferred into C.B.17 scid/scid mice. Recipient mice were treated with human IgG1 Fc (control) or Fc-OPG three times per week in a prophylactic regimen as well as a therapeutic regimen (after 10% body weight loss) and were evaluated for osteopenia and colitis.

Results: Mice that received CD4+CD45RBHi T cells developed osteopenia (as indicated by decreased bone density accompanied by decreased osteoblasts and increased osteoclasts) and colitis (as indicated by histological changes in the large intestine). Mice that received CD4+CD45RBLo T cells developed neither osteopenia nor colitis. All animals consumed, on average, the same amount of food and water over the course of the study. Prophylactic treatment with Fc-OPG increased bone density in mice that received either CD4+CD45RBHi or CD4+CD45RBLo T cells but had no effects on the gastrointestinal tract. Fc-OPG treatment of osteopenic mice with established IBD caused the normalisation of bone density. Osteopenia in CD4+CD45RBHi T cell recipients was accompanied by hypoparathyroidism that was partially normalised by treatment with Fc-OPG. CD4+CD45RBHi T cell recipients also had a bone marrow inflammatory cell infiltrate expressing tumour necrosis factor α which was unaffected by treatment with Fc-OPG.

Conclusions: CD4+CD45RBHi T cell transfer results in osteopenia in addition to colitis. Evidence suggests that this osteopenia was induced by inflammatory cell infiltration and not by malabsorption of calcium. Recombinant human osteoprotegerin effectively treated the osteopenia. OPG may be a useful therapeutic option for treating osteopenia in patients with IBD.

  • ALP, alkaline phosphatase
  • BMD, bone mineral density
  • IBD, inflammatory bowel disease
  • OPG, osteoprotegerin
  • PTH, parathyroid hormone
  • pQCT, peripheral quantitative computer tomography
  • RANK(L), receptor activator of nuclear factor κB (ligand)
  • SAA, serum amyloid A
  • TRAP, tartarate resistant acid phosphatase
  • TNF-α, tumour necrosis factor α
  • inflammatory bowel disease
  • bone
  • osteoporosis
  • osteoprotegerin

Statistics from Altmetric.com

Bone loss and osteoporosis have long been recognised as long term sequelae of inflammatory bowel diseases (IBD).1,2 Many studies have shown a direct link between inflammatory disease and bone loss, both in animal models3–5 and human disease.6–8 However, the relative contributions of inflammation and/or malabsorption to bone mineral density (BMD) loss is still unclear.9–11 Although various therapies have been tried to specifically treat the disease associated bone loss,12–14 there is still an acute need for effective targeted therapy for this extraintestinal manifestation of the disease.

Osteoprotegerin (OPG) is a soluble decoy receptor that can inhibit the differentiation, survival, and function of osteoclasts by binding and neutralising receptor activator of nuclear factor κB ligand (RANKL), thus preventing the latter from stimulating receptor activator of nuclear factor κB (RANK) on osteoclasts.15,16 OPG knockout mice develop severe early onset osteoporosis17,18 whereas OPG transgenic mice develop osteopetrosis.19 Administration of recombinant OPG has been shown to be effective at preventing bone loss in animal models of arthritis,20–22 bone cancer metastasis,23 and oestrogen deficiency associated with ovariectomy.

We sought to determine whether the CD4+CD45RBHi T cell transfer model of IBD was associated with osteopenia. Secondly, we aimed to determine whether recombinant OPG would prevent or reverse any bone loss in this disease model. Finally, we hoped to assess the relative contributions of inflammation and/or malabsorption to any observed bone loss.

MATERIALS AND METHODS

Mice and reagents

CB6F1 mice aged 12–14 weeks from Charles River Laboratories (Worcester, Massachusetts, USA) were used as T cell donors while female C.B.17 scid/scid mice aged 14–16 weeks were used as recipients (Jackson Laboratories Bar Harbour, Maine, USA). The fusion protein IgG1Fc-hOPG was prepared as previously described.24

Induction of IBD-like disease

Disease was induced by intraperitoneal injection of 4×105 sorted CD4+CD45RBHi or CD4+CD45RBLo (control) T cells from CB6F1 mice into immunodeficient C.B.17 scid/scid mice.25 Mice were injected subcutaneously with 5.0 mg/kg of Fc-OPG or 3.4 mg/kg of Fc (the equimolar amount) thrice weekly starting on day 0 after disease induction. A second group of mice (CD4+CD45RBHi) were started on Fc-OPG treatment after 10% body weight loss. Mice were housed individually with food and water depletion monitored on a weekly basis and were sacrificed on day 34. Experiments were performed on groups of 4–10 mice and repeated at least once.

Necropsy

Mice were euthanised by CO2 asphyxiation and blood taken for complete blood counts and to derive serum. Tissue samples were fixed in formalin and later stained with haematoxylin and eosin.25 Tibias and femurs were also isolated and assayed using peripheral quantitative computer tomography (pQCT) and radiography or prepared for histological examination according to standard procedures.19 The thoracic vertebrae, as one unit, were fixed and prepared for histological examination as above.19 The lumbar vertebrae were removed and stored in 70% ethanol: L5 was recognised as the first vertebra next to the iliac crest and subjected to pQCT.

Radiography and bone mineral density measurement

Femurs fixed in 70% ethanol were radiographed with a 43855A x ray system (Faxitron X-ray, Buffalo Grove, Illinois, USA) using exposure times of 49 seconds, and magnified with DeskScan II scanning software (Hewlett Packard, Palo Alto, California, USA). BMD was evaluated by pQCT using tibias stored in 70% ethanol and an XCT-960M scanner (Norland Medical Systems, Ft Atkinson, Wisconsin, USA) and related XMICE software (Stratec, Frankfurt, Germany). Two 1.25 mm cross sections of the proximal metaphysis of the tibia, 1.5 mm and 2 mm, respectively, from the end of the bone, were analysed to determine total and trabecular BMD and their average value calculated. To determine cortical BMD, a single 1.25 mm cross section of the diaphysis of the tibia, 4 mm from the proximal end of the bone, was analysed. Total BMD was also determined analysing the lumbar vertebra L5 and focusing on a single 1.25 mm mid-vertebral cross section.

Bone histomorphometry

Histomorphometric analysis was performed using paraffin embedded sections of the tibias, as previously described.26 Osteoblasts were identified morphologically from sections stained with haematoxylin and osteoclasts by staining for cathepsin K and haematoxylin. Osteoblast and osteoclast perimeter was determined as the perimeter of cells in direct contact with cancellous bone surfaces.

Histological examination of the large intestine and bone

Histopathological damage of the proximal, middle, and distal large intestine was semi quantitatively scored as a global assessment of inflammation on a scale of 0–5.25 Histopathological damage of the bone was assessed by examining tibias and thoracic vertebrae stained with haematoxylin and eosin. Damage was scored semi quantitatively using the aforementioned scoring system, based primarily on inflammatory infiltrate.23 All samples were scored by a pathologist in a blinded fashion.

Serum amyloid A, complete blood counts, and clinical chemistries

Serum amyloid (SAA) was measured in serum by enzyme linked immunosorbent assay using a commercially available kit (BioSource International, Camarillo, California, USA). Complete blood counts were obtained using an H1E counter (Technicon, Terrytown, New York, USA). Serum was analysed for the presence of calcium, phosphorous, and alkaline phosphatase activity using specific diagnostic kits from Roche Diagnostics (New Jersey, USA) on a Hitachi 717 chemistry analyser (Hitachi, New Jersey, USA). Serum tartarate resistant acid phosphatase (TRAP) was analysed on the same machine using kit reagents from Sigma Diagnostics (St Louis, Missouri, USA). Parathyroid hormone (PTH) was measured in serum by radioimmunoassay using a commercially available kit (Immutopics, San Clemente, California, USA).

In situ hybridisation

An antisense RNA probe for a fragment of the murine tumour necrosis factor α (TNF-α nucleotides 429–595 of GenBank accession No M13049.1) was synthesised from a linearised plasmid template with [33P] UTP (Amersham, Piscataway, New Jersey, USA) and the Sp6 RNA polymerase (Ambion, Austin, Texas, USA). Sections (5 μM) of the paraffin embedded, formalin fixed, decalcified tibia were processed by standard in situ hybridisation techniques, as previously described,27 and hybridised overnight with 2×106 counts per minute of probe. TNF-α mRNA expression in the tibia and femur was designated positive or negative, indicating the intensity and extent of signal in the entire section. All samples were scored in a blinded fashion.

Statistical analysis

Results are given as means (SEM) of the data set. Body weight curves were compared using the t statistic in an ANOVA with Dunnett’s correction. Continuous variables such as BMD, and discrete variables such as histopathological scores, were compared using the unpaired heteroschedastic Student’s t test. Variables that were scored as positive or negative (such as presence or absence of TNF-α) were analysed by χ2 analysis.

RESULTS

Transfer of CD4+CD45RBHi T cells to immunodeficient mice causes body weight loss and systemic and colonic inflammation, which are unaffected by treatment with Fc-OPG

Transfer of CD4+CD45RBHi T cells to immunodeficient mice caused significant progressive body weight loss whereas transfer of CD4+CD45RBLo T cells had no significant effect on body weight (p<0.0001) (data not shown). Similar results were seen with elevated levels of SAA or other disease associated haematological abnormalities, including an elevated white blood cell count, depressed red blood cell count, and depressed haemoglobin and haematocrit (data not shown). The colon is the primary site of inflammation in this model and fig 1 clearly shows that transfer of CD4+CD45RBHi T cells to immunodeficient mice caused inflammatory cell infiltrate and histopathological damage to the large intestine (fig 1B), compared with transfer of CD4+CD45RBLo T cells which had no effect (fig 1A). Administration of Fc-OPG had no significant further effect on the large intestine of CD4+CD45RBLo (fig 1C) or CD4+CD45RBHi (fig 1D) mice. Quantification and statistical analysis of all of the experimental groups confirmed this observation (CD4+CD45RBLov CD4+CD45RBHi; p<0.0001) (see fig 2) and also showed that therapeutic administration of Fc-OPG—after the mice had lost 10% of their original body weight—had no significant effect on the histopathological damage to the large intestine caused by transfer of the CD4+CD45RBHi T cells.

Figure 1

 Transfer of CD4+CD45RBHi T cells to scid/scid mice caused significant inflammatory cell infiltrate and histopathological damage to the large intestine (B) compared with transfer of CD4+CD45RBLo T cells which had no significant effect on the histopathology of the large intestine of immunodeficient mice (A). Administration of Fc-osteoprotegerin (OPG) has no significant effect on the large intestine of CD4+CD45RBLo (C) or CD4+CD45RBHi (D) mice. (A) CD4+CD45RBLo cells treated with 3.4 mg/kg of Fc three times per week starting on day 0. (B) CD4+CD45RBHi cells treated with 3.4 mg/kg of Fc three times per week starting on day 0. (C) CD4+CD45RBLo cells treated with 5.0 mg/kg of Fc-OPG control three times per week starting on day 0. (D) CD4+CD45RBHi cells treated with 5.0 mg/kg of Fc-OPG three times per week starting on day 0. A representative haematoxylin and eosin stained biopsy from the middle of the large intestine is shown (see materials and methods) in each case. Note the significant mixed cell type inflammatory infiltrate (*), mucin depletion, and crypt abcessation present in the mucosa of the diseased mice injected with CD4+CD45RBHi cells.

Figure 2

 Transfer of CD4+CD45RBHi T cells (RB Hi) to scid/scid mice caused significant inflammation and histopathological damage to the large intestine (LI) of immunodeficient mice whereas transfer of CD4+CD45RBLo T cells (RB Lo) had no significant effect on the histopathology of the large intestine. Administration of Fc-osteoprotegerin (OPG) had no significant effect on the large intestine of CD4+CD45RBLo or CD4+CD45RBHi mice. Disease was induced by injection of 4×105 sorted CD4+CD45RBHi T cells into scid/scid mice. The negative control mice received 4×105 CD4+CD45RBLo cells. Treatment consisted of subcutaneous injection of 5.0 mg/kg of Fc-OPG or 3.4 mg/kg of Fc control three times per week starting on day 0. For the therapeutic group, treatment with Fc-OPG started when individual animals had attained 10% body weight loss. Treatment was continued in all cases until necropsy on day 34. Haematoxylin and eosin stained biopsies of the proximal, middle, and distal large intestine were assessed for extent of inflammation on a scale of 0–5 (0 = absent, 5 = severe) by a pathologist in a blinded fashion and then averaged. In all cases results represent mean (SEM) values, with n = 7–8 for the CD4+CD45RBHi groups and n = 4 for the CD4+CD45RBLo groups. Statistical significance versus the CD4+CD45RBLo group treated with Fc was calculated by the probability associated with the unpaired heteroschedastic Student’s t test: ***p<0.001.

Hence all of the primary parameters associated with IBD-like disease were unchanged with either prophylactic or therapeutic treatment with Fc-OPG and were within normal values. The lack of any effect of administration of recombinant Fc-OPG on any of the inflammation end points associated with this disease model of IBD is consistent with other animal models of autoimmune disease,20,22 as well as cancer23 and hypercalcaemia.24

CD4+CD45RBHi T cells cause bone density loss, which is prevented or normalised by treatment with Fc-OPG

Standard x rays were taken to visualise any significant alterations in the structure or radio-opacity of the bone. Figure 3 shows that transfer of CD4+CD45RBHi T cells to immunodeficient mice caused a visible decrease in the radio-opacity of the distal femur (fig 3B) compared with that observed with transfer of CD45RBLo T cells (fig 3A). Prophylactic administration of Fc-OPG to control CD4+CD45RBLo mice visibly increased the radio-opacity of the distal femur (fig 3C). Prophylactic Fc-OPG seemed to normalise the radiographic appearance of the distal femur of diseased CD4+CD45RBHi mice (fig 3D). We quantified changes in BMD via pQCT analysis of the tibia. Figure 4 demonstrates that transfer of CD4+CD45RBHi T cells to scid/scid mice caused a highly significant decrease in the total (trabecular plus cortical) BMD of the proximal tibial metaphysis compared with transfer of CD4+CD45RBLo T cells (p<0.0001) (fig 4A). CD4+CD45RBHi T cells also caused a decrease in total BMD of the L5 lumbar vertebra (fig 4B). Administration of Fc-OPG to control CD4+CD45RBLo mice supranormalised total BMD of the tibia (p = 0.0013) but had no significant effect on total BMD of the L5 lumbar vertebra (p = 0.7846). Prophylactic Fc-OPG also supranormalised BMD of the tibia of diseased CD4+CD45RBHi mice (p = 0.0003) and therapeutic administration normalised BMD (p = 0.9594). Fc-OPG had no significant effect on BMD of the L5 lumbar vertebra. In all cases, BMD loss in diseased mice was more pronounced in trabecular bone relative to cortical bone (data not shown). In a separate experiment, tibial BMD loss at the 10% body weight loss point (the point of intervention with Fc-OPG) was found to be similar to that at the time of necropsy (p = 0.3052). For reference, previous experiments have shown that there is no significant difference between total, trabecular, and cortical BMD of CD4+CD45RBLo mice and normal unreconstituted scid/scid mice. These data clearly show that administration of Fc-OPG significantly increases BMD of normal mice and can both prevent disease associated bone loss as well as normalising loss that has already occurred.

Figure 3

 Transfer of CD4+CD45RBHi T cells to scid/scid mice caused an obvious decrease in the radio-opacity of the distal femur (B) compared with transfer of CD4+CD45RBLo T cells (A). Prophylactic administration of Fc-osteoprotegerin (OPG) to the CD4+CD45RBLo mice noticeably increased the radio-opacity of the distal femur (C). Prophylactic Fc-OPG also restored the radio-opacity of the distal femur of diseased CD4+CD45RBHi mice (D). Disease was induced by injection of 4×105 sorted CD4+CD45RBHi T cells into scid/scid mice. Negative control mice received 4×105 CD4+CD45RBLo cells. Treatment consisted of subcutaneous injection of 5.0 mg/kg of Fc-OPG or 3.4 mg/kg of Fc control three times per week starting on day 0. Treatment was continued in all cases until necropsy on day 34. Excised femurs were fixed in 70% ethanol, radiographed with a Faxitron x ray system (Model 43855A Faxitron; X-ray Corp, Buffalo Grove, Illinois, USA) with an exposure time of 49 seconds, and magnified 645% with scanning software (DeskScan II v.2.8; Hewlett Packard, Palo Alto, California, USA).

Figure 4

 Transfer of CD4+CD45RBHi T cells (RB Hi) to scid/scid mice caused a highly significant decrease in total bone mineral density (BMD) of the tibia compared with transfer of CD4+CD45RBLo T cells (RB Lo) (A). CD4+CD45RBHi T cells also caused a decrease in total BMD of the L5 lumbar vertebra (B). Administration of Fc-osteoprotegerin (OPG) to CD4+CD45RBLo mice supranormalised total BMD of the tibia but had no significant effect on total BMD of the lumber vertebra. Prophylactic Fc-OPG also supranormalised total BMD of the tibia of diseased CD4+CD45RBHi mice and therapeutic administration normalised it. Fc-OPG had no significant effect on total BMD of the L5 lumbar vertebra. Disease was induced by injection of 4×105 sorted CD4+CD45RBHi T cells into scid/scid mice. Negative control mice received 4×105 CD4+CD45RBLo cells. Treatment consisted of subcutaneous injection of 5.0 mg/kg of Fc-OPG or 3.4 mg/kg of Fc control three times per week starting on day 0. For the therapeutic group, treatment with Fc-OPG started when individual animals had attained 10% body weight loss. Treatment was continued in all cases until necropsy on day 34. Bone density was analysed by peripheral quantitative computer tomography of two 1.25 mm cross sections of the proximal end of the tibia or the L5 vertebrae proximal to the iliac crest. In all cases results represent mean (SEM) values, with n = 7−8 for the CD4+CD45RBHi groups and n = 4 for the CD4+CD45RBLo groups. Statistical significance versus the CD4+CD45RBLo group treated with Fc was calculated by the probability associated with the unpaired heteroschedastic Student’s t test: *p<0.05, ***p<0.001.

CD4+CD45RBHi T cells decrease osteoblasts and increase osteoclasts, while Fc-OPG eliminates both cell types

Previous studies with OPG have demonstrated its ability to reduce the numbers of osteoclasts in both normal and inflamed bone.20,22 Figure 5 shows that the transfer of CD4+CD45RBHi T cells to immunodeficient mice caused a significant decrease in the number of osteoblasts (p = 0.0236) (fig 5A) and an increase in the number of osteoclasts (p = 0.0387) (fig 5B) in the tibia compared with transfer of CD4+CD45RBLo T cells. Administration of Fc-OPG to control CD4+CD45RBLo mice virtually eliminated both osteoblasts and osteoclasts. Fc-OPG, administered either prophylactically or therapeutically, also eliminated all osteoblasts and osteoclasts from the tibia-femur of diseased CD4+CD45RBHi mice. This observation is consistent with previous studies,20 and the simultaneous elimination of osteoblasts is thought to result from the normal coupling of osteoblast and osteoclast metabolism.

Figure 5

 Transfer of CD4+CD45RBHi T cells (RB Hi) to scid/scid mice caused a significant decrease in the number of osteoblasts (A) and an increase in the number of osteoclasts (B) in the tibia-femur compared with transfer of CD4+CD45RBLo T cells (RB Lo). Administration of Fc-osteoprotegerin (OPG) to control CD4+CD45RBLo mice completely eliminated both osteoblasts and osteoclasts. Fc-OPG, administered either prophylactically or therapeutically, also eliminated all osteoblasts and osteoclasts from the tibia-femur of diseased CD4+CD45RBHi mice. Disease was induced by injection of 4×105 sorted CD4+CD45RBHi T cells into scid/scid mice. Negative control mice received 4×105 CD4+CD45RBLo cells. Treatment consisted of subcutaneous injection of 5.0 mg/kg of Fc-OPG or 3.4 mg/kg of Fc control three times per week starting on day 0. For the therapeutic group, treatment with Fc-OPG started when individual animals had attained 10% body weight loss. Treatment was continued in all cases until necropsy on day 34. Results represent mean (SEM) values, with n = 7−8 for the CD4+CD45RBHi groups and n = 4 for the CD4+CD45RBLo groups. The numbers of osteoclasts and osteoblasts were determined by bone histomorphometry, as described in the materials and methods. Statistical significance versus the CD4+CD45RBLo group treated with Fc was calculated by the probability associated with the unpaired heteroschedastic Student’s t test: *p<0.05.

CD4+CD45RBHi T cells decrease circulating parathyroid hormone, which is partially normalised by treatment with Fc-OPG

Disease related malabsorption may include impairment of calcium absorption, which could lead to hypocalcaemia and a rapid compensatory increase in serum PTH. We therefore measured serum PTH levels mid way through the study as well as at necropsy to look for indirect evidence of impaired calcium absorption. Transfer of CD4+CD45RBHi T cells to immunodeficient mice caused a highly significant decrease in levels of circulating PTH (2.3 (1.5) pg/ml) at the time of necropsy compared with transfer of CD4+CD45RBLo T cells (27.4 (2.2) pg/ml) (p<0.0001). Administration of Fc-OPG to control CD4+CD45RBLo mice decreased levels of circulating PTH (8.8 (7.0) pg/ml) but there was considerable variation in the reduction (p = 0.0495). Prophylactic or therapeutic administration of Fc-OPG partially normalised levels of circulating PTH in diseased CD4+CD45RBHi mice relative to CD4+CD45RBLo mice (prophylactic 15.5 (4.6) pg/ml, p = 0.0385; therapeutic 15.1 (6.9) pg/ml, p = 0.1136).

In a separate experiment, serum PTH was also analysed at the 10% body weight loss time point (approximately day 25, individualised to each mouse) and a similar pattern was observed with the exception that the diseased Fc-OPG therapeutically treated mice had serum PTH levels that were not significantly different from CD4+CD45RBLo control mice (p = 0.6406) (data not shown).

CD4+CD45RBHi T cells do not affect serum calcium or phosphorous but they increase serum TRAP and decrease serum alkaline phosphatase; Fc-OPG normalises TRAP and further decreases alkaline phosphatase

Immunodeficient mice injected with CD4+CD45RBHi cells had similar levels of serum calcium to mice injected with CD4+CD45RBLo cells (p = 0.5666), irrespective of treatment with Fc or Fc-OPG (all within the normal range of 8.25–8.50 g/dl (Amgen Inc., historical data)).

Similarly, CD4+CD45RBHi cells did not cause a change in levels of serum phosphorous (9.2 (0.4) g/dl) compared with control CD4+CD45RBLo cells (9.0 (0.5) g/dl; p = 0.7475). Diseased CD4+CD45RBHi mice treated prophylactically with Fc-OPG had significantly lower serum phosphorus (8.0 (0.4) g/dl; p = 0.0561), as did diseased CD4+CD45RBHi mice treated with therapeutic Fc-OPG (7.6 (0.5) g/dl; p = 0.0421). P levels in these two Fc-OPG treatment groups were not significantly different from each other (p = 0.6984).

CD4+CD45RBHi cells caused a significant rise in levels of serum TRAP (6.00 (0.45) U/l) compared with transfer of CD4+CD45RBLo cells (3.48 (0.3) U/l; p<0.001). CD4+CD45RBLo mice treated with Fc-OPG did not have significantly changed levels of TRAP (5.18 (2.13) U/l; p = 0.4267). Levels of TRAP in diseased mice treated prophylactically with Fc-OPG (4.24 (0.85)) were normalised and were not significantly different to CD4+CD45RBLo normal controls treated with Fc (p = 0.3854). TRAP levels in diseased mice treated with therapeutic Fc-OPG were also normalised (4.48 (0.27)) but were still significantly higher than normal controls treated with Fc (p = 0.0133). This observation is consistent with the known correlation of TRAP levels with osteoclast numbers.28,29

CD4+CD45RBHi mice had significantly lower levels of alkaline phosphatase (ALP) (54.9 (3.5) U/l) compared with control CD4+CD45RBLo mice (83.3 (3.8) U/l; p<0.0001). Control mice treated with Fc-OPG also had significantly depressed levels of ALP (52.3 (3.6) U/l; p<0.0001). Levels of ALP in diseased mice treated with prophylactic Fc-OPG were depressed even further (34.6 (1.3) U/l; p = 0.0002 relative to control CD4+CD45RBLo mice), and similarly in therapeutic Fc-OPG treatment (40.6 (4.6) U/l; p = 0.0002 relative to control CD4+CD45RBLo mice). Serum ALP levels in the two diseased Fc-OPG treatment groups were not significantly different from each other (p = 0.1570). This change in ALP levels is consistent with the known correlation of ALP with osteoblast numbers and is consistent with the histomorphometric analysis. Levels of ALP fall in diseased mice as the numbers of osteoblasts decreases and treatment of diseased mice with Fc-OPG eliminates both osteoclasts and osteoblasts, further lowering the levels of ALP.30–32

CD4+CD45RBHi T cells cause a TNF-α expressing bone inflammatory infiltrate, which is unaffected by treatment with Fc-OPG

Previous studies with OPG in inflammatory models of disease have established that OPG exerts its effect by specifically targeting osteoclasts that mediate bone resorption.20,23 Hence we examined whether there was significant inflammatory cell infiltrate in the bones of diseased mice in this model of IBD. Transfer of CD4+CD45RBHi T cells to scid/scid mice induced an inflammatory cell infiltrate (fig 6B, main picture and inset) and an increase in the numbers of osteoclasts (fig 6B, main picture) compared with transfer of non-pathogenic control CD4+CD45RBLo T cells which had no significant effect. This T cell transfer had no significant effect on the composition of normal bone marrow (fig 6A). Administration of Fc-OPG caused increased amounts of trabecular bone and thickening of the cortical bone in control CD4+CD45RBLo mice (fig 6C). Fc-OPG had a similar effect in diseased CD4+CD45RBHi mice, despite the presence of the inflammatory infiltrate (fig 6D), by greatly reducing the numbers of osteoclasts (fig 6C, 6D inserts). Quantification of the inflammatory infiltrate and statistical analysis of all of the experimental groups confirmed this observation (CD4+CD45RBHiv CD4+CD45RBLo; p<0.0001) (fig 7) and also showed that therapeutic administration of Fc-OPG—after the mice had lost 10% of their original body weight—had no further significant effect on the inflammatory infiltrate of the bone caused by transfer of CD4+CD45RBHi T cells. It is noteworthy that the inflammatory infiltrate of the bone was equally pronounced in both the tibia/femur and thoracic vertebrae. The increased numbers of osteoclasts noted in diseased mice and the decreased numbers evident in Fc-OPG treated mice is consistent with the histomorphometric analysis and levels of TRAP and ALP previously described.

Figure 6

 Transfer of CD4+CD45RBHi T cells to scid/scid mice caused significant inflammatory cell infiltrate (B) and an increase in the numbers of osteoclasts (B, inset) compared with transfer of CD4+CD45RBLo T cells which had no significant effect on the composition of normal bone marrow (A). Administration of Fc-osteoprotegerin (OPG) caused increased amounts of trabecular bone and thickening of the cortical bone in CD4+CD45RBLo mice (C) and had a similar effect in diseased CD4+CD45RBHi mice, despite the presence of the inflammatory infiltrate (D), by greatly reducing the numbers of osteoclasts (C and D inserts). (A) CD4+CD45RBLo cells treated with 3.4 mg/kg of Fc three times per week starting on day 0. (B) CD4+CD45RBHi cells treated with 3.4 mg/kg of Fc three times per week starting on day 0. (C) CD4+CD45RBLo cells treated with 5.0 mg/kg of Fc-OPG control three times per week starting on day 0. (D) CD4+CD45RBHi cells treated with 5.0 mg/kg of Fc-OPG three times per week starting on day 0. Disease was induced by injection of 4×105 sorted CD4+CD45RBHi T cells into scid/scid mice. Negative control mice received 4×105 CD4+CD45RBLo cells. Treatment consisted of subcutaneous injection of 5.0 mg/kg of Fc-OPG or 3.4 mg/kg of Fc control three times per week starting on day 0. Treatment was continued in all cases until necropsy on day 34.

Figure 7

 Transfer of CD4+CD45RBHi T cells (RB Hi) to scid/scid mice caused significant inflammatory cell infiltrate in the bone marrow of scid/scid mice compared with transfer of CD4+CD45RBLo T cells (RB Lo). The effect was equally pronounced in the tibia/femur and thoracic vertebrae. Administration of Fc-osteoprotegerin (OPG) had no significant effect on the infiltrate present in CD4+CD45RBLo mice or diseased CD4+CD45RBHi mice. Disease was induced by injection of 4×105 sorted CD4+CD45RBHi T cells into scid/scid mice. Negative control mice received 4×105 CD4+CD45RBLo cells. Treatment consisted of subcutaneous injection of 5.0 mg/kg of Fc-OPG or 3.4 mg/kg of Fc control three times per week starting on day 0. For the therapeutic group, treatment with Fc-OPG started when individual animals had attained 10% body weight loss. Treatment was continued in all cases until necropsy on day 34. Haematoxylin and eosin stained biopsies of the bones were assessed for extent of inflammation on a scale of 0–5 (0 = normal myelopoeisis, 5 = massive myeloid hyperplasia) by a pathologist in a blinded fashion. In all cases results represent mean (SEM) values with n = 7–8 for the CD4+CD45RBHi groups and n = 4 for the CD4+CD45RBLo groups. Statistical significance versus the vehicle control was calculated by the probability associated with the unpaired heteroschedastic Student’s t test: *p<0.05, ***p<0.001.

Recent work has further defined the mechanism by which inflammation leads to bone loss, specifically the importance of TNF-α in inducing RANKL and subsequent osteoclastogenesis.10,21,33 In an attempt to establish the actions of the inflammatory cells in bone marrow, we examined levels of TNF-α mRNA by in situ hybridisation. Transfer of control CD4+CD45RBHi T cells to scid/scid mice caused significant upregulation of TNF-α mRNA (fig 8B) whereas transfer of CD4+CD45RBLo T cells had no significant effect on expression of TNF-α mRNA in the shaft of the tibia (fig 8A). Administration of Fc-OPG to either control CD4+CD45RBLo mice (fig 8C) or diseased CD4+CD45RBHi mice (fig 8D) had no significant further effect on the observed levels of TNF-α mRNA. Quantification of the inflammatory infiltrate and statistical analysis of all of the experimental groups confirmed this observation in that CD4+CD45RBHi mice had detectable levels of TNF-α mRNA in 6/9 mice and treatment with Fc-OPG, in either prophylactic or therapeutic regimens, did not significantly change this observed penetrance (5/7 and 6/9, respectively). However, this TNF-α infiltrate was significantly different from control CD4+CD45RBLo mice by χ2 analysis (p = 0.0035) which, in contrast, had detectable TNF-α mRNA in 1/6 mice in both the Fc and Fc-OPG treated groups.

Figure 8

 Transfer of CD4+CD45RBHi T cells to scid/scid mice caused significant upregulation of tumour necrosis factor α (TNF-α) mRNA in the shaft of the tibia (B) compared with transfer of CD4+CD45RBLo T cells, which had no significant effect on expression of TNF-α mRNA (A). Administration of Fc-osteoprotegerin (OPG) had no significant effect on expression of TNF-α mRNA in the shaft of the tibia/femur in either CD4+CD45RBLo mice (C) or CD4+CD45RBHi mice (D). (A) CD4+CD45RBLo cells treated with 3.4 mg/kg of Fc three times per week starting on day 0. (B) CD4+CD45RBHi cells treated with 3.4 mg/kg of Fc three times per week starting on day 0. (C) CD4+CD45RBLo cells treated with 5.0 mg/kg of Fc-OPG control three times per week starting on day 0. (D) CD4+CD45RBHi cells treated with 5.0 mg/kg of Fc-OPG three times per week starting on day 0. Disease was induced by injection of 4×105 sorted CD4+CD45RBHi T cells into scid/scid mice. Negative control mice received 4×105 non-pathogenic CD4+CD45RBLo cells. Treatment consisted of subcutaneous injection of 5.0 mg/kg of Fc-OPG or 3.4 mg/kg of Fc control three times per week starting on day 0. Treatment was continued in all cases until necropsy on day 34. In situ hybridisation using a standard murine TNF-α probe was performed, as described in the materials and methods, and scoring was performed by a pathologist in a blinded fashion for the presence or absence of signal.

DISCUSSION

We have shown that induction of a Crohn’s disease like phenotype in mice results in significant loss of BMD. Treatment with recombinant OPG results in prevention of bone loss or its normalisation if administration is delayed until animals are grossly symptomatic with respect to body weight loss. Administration of OPG does not appear to affect any of the parameters relating to the primary disease, such as body weight loss or inflammation of the large intestine. The effects of OPG appear to be targeted solely on bone.

Many investigators have established the prevalence of osteoporosis in the Crohn’s disease population,34,35 with occurrence in the ulcerative colitis population being less well established.11,36 This may be because Crohn’s disease is considered to be a more systemic disease than ulcerative colitis, involving the entire gastrointestinal tract with a longer premorbid course and more extraintestinal manifestations.37 We chose this mouse model of IBD to be representative of the Crohn’s disease variant of IBD38 and we did find significant and reproducible BMD loss. This occurred primarily in trabecular rather than cortical bone. Several theories have been proposed to explain the mechanism of action of BMD loss in Crohn’s disease. As none of the mice in this study received any corticosteroid or other immunosuppressive therapy, BMD loss is most likely a result of the primary disease, which is concordant with most of the published studies on Crohn’s disease associated osteoporosis in humans.11,34,39

The use of an animal model allowed us to address the relative contributions of inflammation and/or malabsorption to BMD loss. Most studies in humans have concluded that nutritional factors, vitamin deficiency, and malabsorption are not major contributing factors to BMD loss.9,11,40 We found in this study that all mice consumed, on average, the same amount of food or water over the course of the study, and hence decreased food intake was not a causative agent. The question of whether or not in vivo malabsorption subsequently occurs in mice is more difficult. Several studies in animals have demonstrated that diets deficient in calcium and phosphorus or protein take months to cause BMD loss,41–44 and our study lasted only five weeks. The decrease in circulating PTH and the presence of a TNF-α elaborating cellular infiltrate in the bone suggests that BMD loss is due primarily to inflammatory processes. It is possible that the inflammation led to degradation and resorption of calcified bone matrix, causing an increase in serum calcium. This increase would cause a rapid decrease in circulating PTH levels which would serve to prevent further increases in serum calcium. This homeostatic mechanism appeared to be effective as serum calcium was unchanged in all of the mice in the study, regardless of disease or treatment state. OPG treatment almost normalised decreased PTH levels, possibly as a consequence of decreased bone destruction and release of less free calcium from bone. The presence of a TNF-α elaborating cellular inflammatory infiltrate in the bones of diseased animals is also consistent with this theory, as several investigators have shown a direct link between levels of secreted TNF-α and RANKL induced osteoclast mediated bone destruction.33,36,45–48 It has further been shown that RANKL can act to maintain inflammation.49 Although the reported data in human studies are inconsistent on this point, this theory of increased bone resorption due to proinflammatory cytokines, accompanied by decreased bone formation, is in broad agreement with the majority of the published studies.11,36 Consistent with this, bone histomorphometry in our study revealed that the diseased animals had more osteoclasts and significantly less osteoblasts compared with normal animals. The observed increases in serum TRAP (marker for osteoclasts) and decrease in serum ALP (marker for osteoblasts) are in agreement with this observation, although future studies to investigate the relative roles of inflammation and/or malabsorption are warranted.

Various therapeutic interventions, ranging from calcium supplementation12 or glucagon-like peptide 2,13 to the use of bisphosphonates14 have been tried to treat the osteoporosis associated with the collective IBD patient population but efficacy has ranged from absent to modest, respectively. There is still an acute need for effective targeted therapy for this extraintestinal manifestation of the disease. The use of OPG in this study provided significant protection against IBD associated BMD loss but did not have any significant effect on body weight loss, levels of SAA, or the disease associated haematological abnormalities or inflammation of the large intestine or the bone itself. This targeted effect of OPG is consistent with the many other preclinical studies performed.20,50–52 When OPG administration was delayed until the diseased animals had lost 10% of their body weight, OPG still normalised BMD of the tibiae. It is important to note that this reversal of osteopenia was probably due to the fact that the animals were young and still undergoing longitudinal bone growth. In these animals, osteoclast inhibition would cause an exaggerated increase in BMD that would not likely be observed in skeletally mature humans. In this study we used ongoing prophylactic and therapeutic administration of OPG at a relatively high dose of 5.0 mg/kg three times per week. Other studies in animal models of autoimmune disease have shown that lower and less frequent doses may be equally effective.20

On a cellular level, we found that OPG completely eliminated any detectable osteoclasts from the bone, even with a therapeutic dosing regimen, as well as eliminating any detectable osteoblasts. Elimination of osteoclasts was expected and elimination of osteoblasts is probably due to the tightly coupled regulation of these two cell types. Normalisation of serum TRAP by treatment with Fc-OPG reflects its linkage to levels of osteoclasts and the decrease in serum ALP by treatment with Fc-OPG is consistent with elimination of osteoblasts in bone.

In summary, this study suggests that inflammation is the primary cause of bone loss in IBD and that recombinant human osteoprotegerin may be of significant benefit in the clinical management of IBD associated osteopenia and osteoporosis.

Acknowledgments

We are grateful to Jorge Arias and Jenny Smith of Amgen Inc., Department of Laboratory Animal Resources, for excellent technical assistance and advice, and to Yan Cheng, Annie Luo, and Darlene Kratavil of Amgen Inc., Department of Pathology and Pharmacology, for excellent sample histology processing. We are also grateful to Tom Graves of Amgen Inc., Department of Biostatistics, for assistance with statistical analysis.

REFERENCES

View Abstract

Footnotes

  • Conflict of interest: None declared.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Digest
    Robin Spiller
  • Letter
    A J Ashcroft S R Carding
  • CORRECTION
    BMJ Publishing Group Ltd and British Society of Gastroenterology