SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Objective

Chronic T cell activation is central to the etiology of rheumatoid arthritis (RA), an inflammatory autoimmune disease that leads to severe focal bone erosions and generalized systemic osteoporosis. Previous studies have shown novel cytokine-like activities in medium containing activated T cells, characterized by potent induction of the osteoblastic production of interleukin-6 (IL-6), an inflammatory cytokine and stimulator of osteoclastogenesis, as well as induction of an activity that directly stimulates osteoclast formation in a manner independent of the key osteoclastogenic cytokine RANKL. This study was undertaken to identify the factors secreted by T cells that are responsible for these activities.

Methods

Human T cells were activated using anti-human CD3 and anti-human CD28 antibodies for 72 hours in AIM V serum-free medium to obtain T cell–conditioned medium, followed by concentration and fractionation of the medium by fast-protein liquid chromatography. Biologically active fractions were resolved using sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Major bands were analyzed by mass spectrometry, and a major candidate protein was identified. This novel cytokine was cloned, and its expression was analyzed using recombinant DNA technologies.

Results

A single novel cytokine that could induce both osteoblastic IL-6 production and functional osteoclast formation in the absence of osteoblasts or RANKL and that was insensitive to the effects of the RANKL inhibitor osteoprotegerin was identified in the activated T cell–conditioned medium; this cytokine was designated secreted osteoclastogenic factor of activated T cells (SOFAT). Further analysis of SOFAT revealed that it was derived from an unusual messenger RNA splice variant coded by the threonine synthase–like 2 gene homolog, which is a conserved gene remnant coding for threonine synthase, an enzyme that functions only in microorganisms and plants.

Conclusion

SOFAT may act to exacerbate inflammation and/or bone turnover under inflammatory conditions such as RA or periodontitis and in conditions of estrogen deficiency.

Rheumatoid arthritis (RA) is a chronic inflammatory disease with a complex etiology. Juxtaarticular bone loss occurring around the inflamed joints and generalized systemic bone loss are common features of RA (for review, see refs.1–3). One of the main characteristics of RA is a dense lymphoid cell infiltration into the synovial membrane. Activated T cells are now considered to be potent modulators of bone turnover and are a key source of osteoclastogenic cytokines under inflammatory conditions such as RA (4, 5) and periodontitis (6, 7) and in estrogen deficiency (8–11). We recently reported that activated T cells secrete cytokines that potently stimulate the differentiation of human bone marrow stromal cells into osteoblasts (12, 13) and also secrete an unknown factor that is capable of stimulating the production of interleukin-6 (IL-6) by osteoblasts (14). IL-6 is an osteoclastogenic factor that has been implicated in the bone destruction associated with estrogen deficiency in humans (15–17) and in mice (18, 19) and in the inflammation and osteoporosis associated with RA (14, 16, 20).

In addition, activated T cells have long been known to stimulate osteoclast formation (21–24). T cell–derived production of tumor necrosis factor α (TNFα) has been reported to play a critical role in ovariectomy-induced bone loss in mice (25), and T cell–derived RANKL is reported to be relevant in animal models of RA (26). We and other investigators (27–30) have reported that activated T cells stimulate osteoclastogenesis in vitro by secretion of RANKL. Interestingly, our observations prompted the further controversial finding that activated T cells also significantly induce osteoclast formation by a mechanism that is independent of RANKL, since saturating concentrations of the RANKL inhibitor osteoprotegerin (OPG) failed to neutralize more than 30% of the observed osteoclast formation induced by activated T cells (28).

In the present study, with the use of sequential biochemical purification, mass spectrometry, and recombinant DNA technologies, we identified and analyzed the expression of a novel activated human T cell–secreted cytokine, herein referred to as secreted osteoclastogenic factor of activated T cells (SOFAT). This single cytokine was observed to elicit RANKL- and osteoblast-independent osteoclast formation in an OPG-insensitive manner and also to stimulate IL-6 production by osteoblasts.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Materials.

Antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA), unless indicated otherwise. All other reagents were purchased from Sigma-Aldrich (St. Louis, MO), unless indicated otherwise.

Biochemical purification of SOFAT.

T cell–conditioned medium was collected and concentrated using a 20-ml Amicon Centricon concentrator (Millipore, Bedford, MA). The concentrate was buffer-exchanged to Tris HCl, pH 8.0, and applied to a DEAE–Sepharose column using a fast-protein liquid chromatography (FPLC) system (Invitrogen, Carlsbad, CA). After washing the column with Tris HCl, pH 8.0, the column was eluted with an NaCl gradient of 0–1M. One-milliliter fractions were collected, and aliquots were assayed for IL-6 activity on human osteoblasts using an Endogen enzyme-linked immunosorbent assay (ELISA) (Pierce, Rockford, IL).

Osteoclast formation was resolved by staining the cells with tartrate-resistant acid phosphatase (TRAP). The active fractions were then concentrated using an Amicon Centricon concentrator with a molecular weight cutoff of 5,000 daltons (Millipore), and the concentrated protein buffer was exchanged to Tris–saline, pH 7.4. The protein was applied to an FPLC Superdex-200 gel-filtration column in Tris–saline, pH 7.4. One-milliliter fractions were collected, and aliquots were assayed again for IL-6 activity on human osteoblasts and also for osteoclast formation on human monocytes.

Identification of SOFAT.

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE).

Aliquots (5 μg) of purified protein from the DEAE–Sepharose column or the FPLC Superdex-200 column were diluted 3:1 (volume:volume) with 4× Laemmli sample buffer, boiled for 5 minutes, and then separated on a 4–12% Novex Tris–glycine Readymade gel (Invitrogen). The gels were then stained for protein detection with colloidal Coomassie blue stain (Genosystems, Woodlands, TX), according to the manufacturer's instructions.

Mass spectrometry

To identify the purified secreted T cell protein, Coomassie blue–stained bands were cut from 10% SDS-PAGE gels and digested in gel with purified trypsin. The gel was extracted with acetonitrile/trifluoroacetic acid and subjected to gas chromatography and matrix-assisted laser desorption ionization–time-of-flight (MALDI-TOF) mass spectrometry (performed in the Proteomics Center at Washington University). Recombinant human SOFAT (rHuSOFAT) was processed similarly and subjected to MALDI-TOF mass spectrometry. Peaks were identified using ProteinProspector (version 5.2.2; http://www.prospector.ucsf.edu).

Cloning of SOFAT.

T cells were activated with anti-CD3 and anti-CD28 antibodies in AIM V medium in a 5% CO2 incubator at 37°C. After 72 hours, total RNA was extracted using an RNeasy kit (Qiagen, Valencia, CA). The isolated RNA was subjected to reverse transcription–polymerase chain reaction (RT-PCR) using a Qiagen one-step RT-PCR kit. Total RNA (1 μg) was subjected to RT for 30 minutes at 50°C and then heated at 95°C for 15 minutes to destroy the RT enzyme. PCR was conducted at 94°C for 30 seconds, at 60°C for 30 seconds, and at 72°C for 1 minute for 35 cycles. Primers were reconstructed from the nucleotide sequence encoding a hypothetical threonine synthase–like protein that was identified using the National Center for Biotechnology Information (NCBI) protein database with the BLAST search tool, which revealed a match with the DNA sequence of FLJ10916 (GenBank accession no. AK001778). The primers contained flanking att B1 and att B2 recombination sites to allow for rapid recombination-mediated transfer of SOFAT complementary DNA (cDNA) from Gateway pDONOR221 plasmid to bacterial (pDEST17) and mammalian (pDEST26) expression plasmids (Invitrogen). These plasmids encode N-terminal 6X-histidine tags fused to the recombinant protein to facilitate protein identification and purification. The primers for detection of SOFAT were as follows: forward 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTCCATGGACATTATCGTTCTGCTGCCC-3′, and reverse 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTTCTACTGGGAGGTGTTGAGGGCATG-3′ (underline indicates the SOFAT sequence). We used Escherichia coli–generated His-tagged rHuSOFAT for all experiments, unless indicated otherwise. The pDEST26 plasmid was used to express N-terminal His-tagged rHuSOFAT in CHO cells to investigate secretion of SOFAT in a mammalian system.

Induction of rHuSOFAT expression.

The SOFAT cDNA in pDEST17 was used to express a recombinant SOFAT protein in E coli BL21-A1. After a 3-hour induction with arabinose, the bacteria were harvested and lysed, and the protein was extracted from inclusion bodies and refolded using the Novagen protein-refolding kit (EMD Biosciences, Darmstadt, Germany) according to the manufacturer's protocol. After refolding, the protein was further purified on a Qiagen Ni-NTA column according to the procedures recommended by the manufacturer.

Isolation of human T cells and monocytes.

Peripheral blood mononuclear cells were obtained from buffy coats (American Red Cross, St. Louis, MO) and were further purified by separation on Histopaque (1.077 gm/ml) lymphocyte-separation medium. T cells were immunomagnetically isolated as previously described (31). The use of human buffy coats was approved by the Washington University Institutional Review Board. CD14+ monocytes were purchased from Stem Cell Technologies (Vancouver, British Columbia, Canada) or were isolated from buffy coats using EasySep CD14 Positive Selection Cocktail and EasySep Magnetic Nanoparticles (Stem Cell Technologies).

Analysis of osteoclast formation.

Human monocytes.

Purified human monocytes were plated at a density of 5 × 105 cells/well in 48-well plates in a final volume of 0.5 ml α-minimum essential medium (α-MEM) supplemented with 10% fetal bovine serum (FBS) (Hyclone, Logan, UT), penicillin (100 units/ml), streptomycin (100 μg/ml), and 25 ng/ml recombinant human macrophage colony-stimulating factor (M-CSF) (R&D Systems, Minneapolis, MN). Cultures were stimulated with either 10% purified T cell SOFAT, various doses of rHuSOFAT, or 25 ng/ml recombinant RANKL (a kind gift from Dr. Beth Lee, Ohio State University, Columbus, OH). To test for RANKL-dependent versus RANKL-independent osteoclast formation, cultures were treated with 200-fold excess OPG (R&D Systems) relative to RANKL. The medium (50%) was changed every 3–4 days for 10 days, and then the cells were analyzed for osteoclast formation by TRAP staining using a leukocyte acid phosphatase kit.

Murine RAW 264.7 cells.

Osteoclasts were generated from the mouse monocytic cell line RAW 264.7 (American Type Culture Collection, Manassas, VA) using recombinant human RANKL (rHuRANKL) (60 ng/ml) and crosslinking anti–6X-His antibody (2.5 μg/ml) (R&D Systems) or rHuSOFAT (100 ng/ml). RAW 264.7 cells were seeded into 96-well plates (5–10,000 cells/well) in a final volume of 200 μl of α-MEM supplemented with 10% FBS (Hyclone), penicillin (100 units/ml), and streptomycin (100 μg/ml). RAW 264.7 cells were cultured at 37°C in a 5% CO2 incubator for 5–7 days, followed by TRAP staining for detection of osteoclast formation. In some experiments, cultures also received cyclosporin A (CSA) (2 μg/ml), FK-506 (10 ng/ml), neutralizing TNFα antibody (20 μg/ml), or IL-6 antibodies (20 μg/ml) (AF-410-NA and AB-406-NA; R&D Systems). Osteoclasts were quantitated under light microscopy, with values normalized for the number of nuclei. TRAP-positive cells with ≥3 nuclei were defined as osteoclasts.

Immunocytochemical and microscopy analyses of osteoclasts.

Immunocytochemistry was performed as previously described (27). Briefly, multinucleated cells were generated using rHuSOFAT (100 ng/ml), as described above, followed by fixation of the cells for 60 seconds with acetone/methanol (50:50 volume:volume) and incubation overnight at 4°C in 3% bovine serum albumin containing specific mouse or goat anti-human IgG antibodies against αv and β3 integrin subunits, cathepsin K, or the osteoclast-specific antibody 121F (a generous gift from Dr. Philip Osdoby, Washington University). Nonspecific binding was assessed by isotype and species (mouse or goat)–matched control antibodies. After 2 washes in phosphate buffered saline (PBS), cells were incubated with secondary antibody (anti-goat or anti-mouse conjugated with horseradish peroxidase) for 2 hours at room temperature. After 3 washes in PBS, color was developed using 4-chloronapthol (0.03% in 0.05M Tris HCI, pH 7.6, and 0.1% H2O2).

Actin ring formation was visualized by fluorescence microscopy on cells stained with phycoerythrin-conjugated anti–filamentous actin antibody. Fluorescence and light microscopy were performed on a Nikon Eclipse TE2000-S inverted phase-contrast microscope. Images were captured using a digital camera (QImaging, Burnaby, British Columbia, Canada).

Osteoclast activity assay.

Resorption was assessed on BD BioCoat Osteologic films (BD Biosciences, Franklin Lakes, NJ), an artificial resorbable calcium phosphate film coated onto a quartz substrate. Osteoclasts were cultured directly from RAW 264.7 cells that were seeded on the BioCoat Osteologic films and generated with rHuSOFAT or rHuRANKL. After 10 days in culture, cells were dissociated with 6% bleach for 10 minutes, followed by 3 washes in water. Resorption pits were digitally photographed under brightfield and phase-contrast microscopy at 200× magnification.

T cell and osteoblast cultures.

T cells were cultured at 1 × 106 cells/ml in AIM V serum-free medium (Gibco, Grand Island, NY). The T cells were then activated by a method as previously described (13, 31), using anti-human CD3 and anti-human CD28 antibodies (PharMingen, San Diego, CA). After a 72-hour incubation period at 37°C, the activated T cell–conditioned medium was harvested and stored at −80°C. Human osteoblasts were prepared from human ribs as previously described (14, 31). The rib specimens were obtained from deceased organ transplant donors through Midwest Transplant Services (St. Louis, MO).

Statistical analysis.

Statistically significant differences were determined using SigmaStat software (Systat, Richmond, CA). Data are expressed as the mean ± SEM or mean ± SD, as appropriate. Group mean values were compared either by one-way analysis of variance using Fisher's least significant difference test or the Tukey-Kramer post hoc test or by the Mann-Whitney test for nonparametric data, as appropriate. P values less than or equal to 0.05 were considered significant.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Isolation and identification of a novel T cell cytokine with osteoclastogenic and osteoblastic IL-6–inducing activities.

To identify the factors responsible for osteoblastic IL-6 production and RANKL-independent osteoclast formation, activated T cell–conditioned medium was subjected to biochemical fractionation by DEAE–Sepharose anion-exchange chromatography, and proteins were eluted with an NaCl gradient (Figure 1A). Fractions were assayed for IL-6 induction in human osteoblasts, and for osteoclastogenic activity on human monocyte cultures pretreated with 25 ng/ml M-CSF (Figures 1A and B). A major peak of IL-6–inducing activity was eluted early in the salt gradient (Figure 1A), and interestingly, this also corresponded with the potency of osteoclastogenic activity (Figure 1B).

thumbnail image

Figure 1. Induction of interleukin-6 (IL-6) and tartrate-resistant acid phosphatase–positive (TRAP+) osteoclast-like cells by fractionated activated T cell–conditioned medium. A and B, DEAE–Sepharose anion-exchange chromatography was used to resolve peak fractions for the assay of IL-6 production by osteoblasts eluted on an NaCl gradient (A) or osteoclastogenic activity on human monocyte cultures indicated by multinucleated TRAP+ cells (representing osteoclasts) (B). Control = untreated cells; Input = unfractionated T cell–conditioned medium. Broken line in A indicates the NaCl gradient. C and D, Superdex-200 fractionation was used to further assess the major DEAE–Sepharose peak (fractions 21–29 in A, with the major peak protein also indicated by the broken line in C), with results revealing a single peak containing both IL-6 activity (C) and osteoclastogenic activity (D). ODA280 = optical density at an absorbance of 280 nm.

Download figure to PowerPoint

Fractions comprising the major peak eluted from the DEAE–Sepharose column (fractions 21–29 in Figure 1A) were pooled and further subjected to Superdex-200 gel-filtration chromatography (Figures 1C and D). The eluted fractions were again assayed for osteoblastic IL-6 activity and also for osteoclast-inducing activity. A single peak, with an apparent molecular mass of ∼27 kd, induced IL-6 production from human osteoblasts (Figure 1C) and again induced osteoclastogenic activity when added to purified human monocytes (Figure 1D).

The peak IL-6– and osteoclast-inducing fractions from the Superdex-200 column (fractions 32–34 in Figure 1C) were again pooled and then subjected to analysis by SDS-PAGE. A major, ∼27-kd product was excised, digested with trypsin, and evaluated by mass spectrometry. Four peptides were identified, and the amino acid sequences were used to interrogate the NCBI protein database using the BLAST tool (32). BLAST analysis revealed strong amino acid sequence homology with a hypothetical threonine synthase–like protein, based on an open-reading frame (GenBank accession no. AK001778) identified by shotgun genomic cDNA sequencing analysis (FLJ10916).

Based on the results of mass spectrometry and matching of the cDNA sequence reported for FLJ10916, we designed RT-PCR primers (see Materials and Methods) and performed RT-PCR on resting and activated T cell total RNA. Activated T cells, but not resting T cells, expressed messenger RNA (mRNA) for the novel protein as well as for RANKL (details available from the corresponding author upon request). Based on the potent osteoclastogenic activity of this factor, we have named it secreted osteoclastogenic factor of activated T cells, or SOFAT.

Nucleotide and deduced amino acid sequence of SOFAT.

The SOFAT cDNA was cloned as described in Materials and Methods. The full DNA sequence was determined by automated sequencing, which revealed that the clone had a sequence somewhat different from that of the previously predicted hypothetical cDNA FLJ10916. The 1,002-bp sequence was found to contain a stop codon at nucleotides 742–744 and translated a protein sequence containing 247 amino acids, identical to the natural product isolated biochemically. The amino acid sequence was deduced from SOFAT cDNA using Vector NTI (Invitrogen), as shown in Figure 2A.

thumbnail image

Figure 2. Nucleotide sequence, protein translation, and gene structure of human secreted osteoclastogenic factor of activated T cells (SOFAT). A, Human SOFAT was cloned based on the alignment of the amino acid sequence derived from mass spectrometry with that for the hypothetical protein FLJ10916, and the nucleotide sequence was thus deduced. The positions of the 3′ and 5′ reverse transcription–polymerase chain reaction primers are underlined. The nucleotide sequence of human SOFAT was determined by sequencing of cDNA reverse transcribed from activated T cell total RNA, and the protein sequence was thus derived. Human SOFAT comprises 247 amino acids (a larger version of A is shown in Supplementary Figure 2, available at the Arthritis & Rheumatism Web site at http://www3.interscience.wiley.com/journal/76509746/home). B, The SOFAT splice variant was derived from the threonine synthase–like 2 (THNSL2) splice variant b reference sequence (details available from the corresponding author upon request). Exons are shown as numbered black boxes, introns as lines (not to scale), and untranslated regions as narrow open bars. SOFAT shows highest homology with splice variant b, and comprises the last 97 bp of exon 4 joined to exons 5, 6, and 7, plus the first 141 bp of exon 8. An additional 33 bp of the 3′-untranslated region (3′-UTR) sequence and at least 258 bp of the noncoding UTR derived from an unused exon 9 is contained within the boundaries of the 3′ primer.

Download figure to PowerPoint

Genomic structure of threonine synthase–like 2 (THNSL2) and mRNA splice variants giving rise to SOFAT.

The SOFAT nucleotide sequence was then compared against the NCBI human genome database using BLAST and was found to have significant homology with parts of the THNSL2 gene located on chromosome 2 (2p11.2). THNSL2 is a gene remnant that encodes a protein highly similar to threonine synthase but is incapable of L-threonine biosynthesis in vertebrates (33). Bioinformatic analysis using the 2007 human version of NCBI AceView (available at http://www.ncbi.nlm.nih.gov/IEB/Research/Acembly/) (34) revealed that the THNSL2 gene contains 26 different introns, and that 13 different mRNA are transcribed. Of these mRNA, 12 are alternatively spliced variants (designated a through l), and the other is an unspliced form (designated m). The mRNA appear to differ based on truncation of the 5′ and/or 3′ ends, the presence or absence of 13 exons, overlapping exons with different boundaries, and alternative splicing or retention of 10 introns.

Interestingly, the sequence of SOFAT represents a fourteenth alternative splice variant that has significant, but not complete, homology with the variant designated b. We found that SOFAT comprises the last 97 of 153 nucleotides of exon 4, spliced to 231 bp of exon 5, 149 bp of exon 6, 126 bp of exon 7, and 141 bp of exon 8, as determined using THNSL2 splice variant b as the reference sequence (Figure 2B). The SOFAT mRNA also contains an additional 33 bp of noncoding 3′-untranslated region (3′-UTR) derived from the intronic sequence immediately following exon 8, and at least (within the boundaries of our 3′ primer) 258 bp of additional 3′-UTR derived from a ninth exon. Exon 9 is noncoding in both THNSL2 variant b and SOFAT but is encoding in several other variants (a to l) (for details on the splice variants, see the compact gene diagram at the Web site http://www.ncbi.nlm.nih.gov/IEB/Research/Acembly/av.cgi?exdb=AceVice&db=36a&term=FLJ10916&unit=Go).

A number of splice variants coded by the THNSL2 gene contain a pyroxidal-5′-phosphate–dependent β-subunit domain. However, SOFAT does not contain the entire domain, suggesting that unlike threonine synthase, SOFAT is not pyroxidal-5′-phosphate dependent.

Our RT-PCR primers had the capacity to identify 7 of the 14 possible variants (SOFAT, as well as variants a, b, d, e, f, and g). However, we routinely observed only 1 major product, SOFAT, in activated T cells, although the region between the primers was the same for SOFAT and variant b and was only 19 bp and 12 bp shorter for variants a and f, respectively (details available from the corresponding author upon request). Consequently, these variants could not be readily resolved from each other, and the extent to which each of these 4 splice isoforms contributes to SOFAT generation remains to be determined.

Expression of recombinant SOFAT.

Recombinant SOFAT was expressed from pDEST17 in E coli BL21-A1, and the resulting His-tagged protein was resolved by SDS-PAGE and Western blotting, using a goat antipolyhistidine antibody (Figure 3A). The protein was found to have a molecular weight of ∼27 kd, identical to that of the natural protein. Mass spectrometry analysis of the purified protein (Figure 3B) revealed a sequence identical to the predicted amino acid translation derived from the cDNA (Figure 2A), thus confirming the identity of the recombinant protein as SOFAT.

thumbnail image

Figure 3. Induction of expression of recombinant human secreted osteoclastogenic factor of activated T cells (rhSOFAT), protein translation, and interleukin-6 (IL-6)–inducing activity of SOFAT. A,Escherichia coli–expressed rhSOFAT was resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and Coomassie blue staining (lane 1) and immunoprobed with an antipolyhistidine antibody (lane 2). B, The identity of the secreted protein was verified by sequencing with mass spectrometry. C, To confirm its secretion, SOFAT was immunoprecipitated from CHO cell supernatants and assayed by Western blotting with an antipolyhistidine antibody. Replicate assays of rhSOFAT are shown in lanes 2 and 3; empy vector was used as a control. D, Induction of IL-6 production in osteoblasts by various doses of rhSOFAT (in comparison with untreated and bovine serum albumin (BSA)–treated controls) was analyzed by enzyme-linked immunosorbent assay. Bars show the mean and SEM results from triplicate cultures. ∗ = P < 0.001 versus untreated and BSA-treated controls, by one-way analysis of variance.

Download figure to PowerPoint

SOFAT was originally identified in the T cell–conditioned medium. However, sequence analysis using SignalP software (version 3.0) (35, 36) for prediction of classically secreted proteins failed to identify any signal peptide in SOFAT. In order to verify that SOFAT is secreted by mammalian cells, the cDNA for SOFAT was transferred to the mammalian expression vector pDEST26, and recombinant His–tagged SOFAT was expressed in CHO cells. The conditioned medium was immunoprecipitated with an antipolyhistidine antibody, and the immunoprecipitated protein was subjected to Western blotting and His-tag immunoprobing. Empty vector–transfected cultures were processed in the same manner. SOFAT was reproducibly produced and secreted from mammalian cells (Figure 3C, second and third lanes). We found no bands in control cultures (Figure 3C, empty vector lane) or in cell lysates from SOFAT-expressing cells (results not shown), suggesting that the protein is rapidly secreted.

Induction of IL-6 production by human osteoblasts and osteoclast formation by purified human monocytes with rHuSOFAT.

Since the natural form of SOFAT induces the secretion of IL-6 in human osteoblasts, we tested rHuSOFAT for this activity. Recombinant human SOFAT induced IL-6 production in human osteoblasts in a dose-dependent manner (Figure 3D), confirming that the recombinant protein has the expected biologic activity on osteoblasts.

Similarly, we confirmed that treatment of purified human CD14+ monocytes with rHuSOFAT resulted in the formation of TRAP+ multinuclear cells in the absence of exogenous RANKL or osteoblasts (Figure 4A). This effect of rHuSOFAT again occurred in a dose-dependent manner, with doses as low as 6 ng/ml stimulating significant osteoclast formation and a maximal production of osteoclasts being induced at a dose of 50 ng/ml.

thumbnail image

Figure 4. Generation and characterization of human osteoclasts induced by recombinant human secreted osteoclastogenic factor of activated T cells (rhSOFAT). A, To evaluate the dose response of human monocytes to rhSOFAT, cells were cultured with various doses of rhSOFAT for 10 days and stained for osteoclast formation using tartrate-resistant acid phosphatase (TRAP). B, TRAP+ multinucleated cells were validated as osteoclasts with the use of multiple specific markers of the osteoclast phenotype. Monocytes were cultured with rhSOFAT (100 ng/ml) for 7 days and immunostained with mouse IgG antibodies against the αv and β3 integrin subunits, with the antiosteoclast antibody 121F, or with goat IgG antibodies against cathepsin K; relevant mouse or goat IgG isotypes were used as controls. In addition, actin ring formation was visualized under fluorescence microscopy. C, Human monocytes were cultured with either 25 ng/ml RANKL or 100 ng/ml rhSOFAT without or with 200-fold excess osteoprotegerin (OPG) for 10 days and evaluated for osteoclast formation by TRAP staining. (Original magnification × 100 in A–C, except as indicated in parentheses in B.)

Download figure to PowerPoint

Characterization of osteoclasts.

The TRAP+ multinucleated cells generated by rHuSOFAT were validated as osteoclasts with the use of multiple specific markers of the osteoclast phenotype. Cells stained positive for αv and β3 integrin subunit expression exhibited reactivity to the antiosteoclast antibody 121F (37) and cathepsin K and displayed formation of actin rings (Figure 4B).

Induction of osteoclast formation by rHuSOFAT in human monocytes in a RANKL-independent manner.

Since SOFAT induces osteoclastogenesis in a manner independent of exogenously added RANKL, we examined whether SOFAT may act by inducing autologous expression of RANKL by human monocytes. When human monocytes were treated with RANKL, osteoclast formation occurred, as expected. This process was inhibited by the simultaneous presence of 200-fold excess OPG (Figure 4C). The addition of rHuSOFAT to human monocytes likewise induced osteoclast formation, but the addition of OPG did not abrogate the effect of rHuSOFAT, demonstrating that the mechanism of SOFAT induction of osteoclastogenesis is not mediated by RANKL (Figure 4C).

Induction of osteoclast formation by rHuSOFAT in the murine system.

The sequence of human SOFAT was found to be aligned to human THNSL2 variant b as well as to multiple eukaryotic sequences, as determined using the BLAST search routine to interrogate the NCBI protein sequence database (see Supplementary Figure 1, available at the Arthritis & Rheumatism Web site at http://www3.interscience.wiley.com/journal/76509746/home). The THNSL2 amino acid sequence coding for SOFAT has been highly evolutionarily conserved across species as diverse as the sea urchin and cow, despite the inability of this protein to synthesize L-threonine (33). We observed strong homologies between humans and other species, including other primates such as orangutan (100% homology) and chimpanzee (98.8% homology), and a high degree of core homology between human SOFAT and the predicted mouse sequence (87% homology).

To investigate whether, as with most other cytokines, this human cytokine is active in the mouse, which represents a model that offers significant advantages in terms of experimental manipulation, we assessed the effect of rHuSOFAT on osteoclast formation by RAW 264.7 cells. RAW 264.7 cells are a well-characterized mouse clonal monocytic cell line that generates osteoclasts at high frequency in response to RANKL. Treatment with rHuSOFAT potently induced TRAP+ multinucleated cell formation by RAW 264.7 cells (Figure 5A).

thumbnail image

Figure 5. Induction of functional osteoclasts in mouse RAW 264.7 cells by recombinant human secreted osteoclastogenic factor of activated T cells (rhSOFAT). A, Mouse RAW 264.7 cells were left unstimulated or were treated with rhSOFAT (100 ng/ml) for 5 days and evaluated for osteoclast formation by tartrate-resistant acid phosphatase staining (original magnification × 100). B, To test murine osteoclast activity, RAW 264.7 cells were left unstimulated or were treated with rhSOFAT (100 ng/ml) or RANKL (60 ng/ml) and cultured on BioCoat Osteologic films for 10 days. Top, Phase-contrast microscopy of resorption pits (original magnification × 200). Bottom, Brightfield microscopy of pits (original magnification × 100). Color figure can be viewed in the online issue, which is available at http://www.arthritisrheum.org.

Download figure to PowerPoint

Induction of functional osteoclasts by SOFAT.

To test whether the TRAP+ multinucleated cells generated by rHuSOFAT are functional osteoclasts, we seeded RAW 264.7 cells on BioCoat Osteologic films, generated osteoclasts using rHuSOFAT or rHuRANKL, and then 10 days later, photographed resorption pits under brightfield and phase-contrast microscopy. Both RANKL- and SOFAT-induced osteoclasts caused significant pit formation on the BioCoat, thus validating these cells as functional osteoclasts (Figure 5B).

Potent amplification of rHuSOFAT-induced osteoclastogenesis by TNFα, but without dependence on TNFα.

TNFα is an inflammatory cytokine that is found at high levels in the inflammatory pannus in RA and is central to the etiology of RA. Although it does not stimulate osteoclast formation directly, TNFα has been shown to potently amplify RANKL-induced osteoclast formation (9). Similarly, our results demonstrated that TNFα also potently amplified rHuSOFAT-induced osteoclast formation (Figures 6A and D), whereas depletion of TNFα from the system using a neutralizing antibody failed to prevent the osteoclast formation induced by rHuSOFAT (Figures 6B and D).

thumbnail image

Figure 6. Induction of osteoclast formation by recombinant human secreted osteoclastogenic factor of activated T cells (rhSOFAT) occurs in a manner independent of tumor necrosis factor α (TNFα) and interleukin-6 (IL-6), despite the capacity of TNFα to amplify rhSOFAT-induced osteoclastogenesis. AC, RAW 264.7 cells were cultured with rhSOFAT (100 ng/ml) without or with recombinant TNFα (rTNFα) (10 ng/ml) (A), neutralizing anti-TNFα antibody (Ab) (20 μg/ml) (B), or anti–IL-6 antibody (20 μg/ml) (C). After 5 days, tartrate-resistant acid phosphatase–positive (TRAP+) multinucleated cells were quantitated under light microscopy. Bars show the mean and SD results from 6 replicate wells. ∗ = P < 0.001 versus unstimulated cells; ∗∗ = P < 0.001 versus rhSOFAT-treated cells, by one-way analysis of variance (Tukey-Kramer test). D, A representative field of TRAP-stained cells is shown for each experiment, with the various treatments compared with unstimulated cells (original magnification × 100, except as indicated in parentheses). In a pilot experiment, stimulation with TNFα and treatment with anti-TNFα and anti–IL-6 antibodies were observed to have no effect on RAW 264.7 cells in the absence of SOFAT (results not shown). We also verified that the TNFα antibody that we used was able to neutralize the effect of 10 ng/ml of rTNFα in the induction of osteoclast formation (results not shown). Color figure can be viewed in the online issue, which is available at http://www.arthritisrheum.org.

Download figure to PowerPoint

Failure of rHuSOFAT to induce IL-6 production by macrophages.

IL-6 is another inflammatory cytokine commonly associated with RA and has been reported to support osteoclast formation. As noted above, we demonstrated that rHuSOFAT stimulates potent IL-6 secretion by osteoblasts. To determine whether SOFAT-induced osteoclast formation also involves, in part, direct IL-6 secretion from osteoclast precursors, we added a neutralizing IL-6 antibody to rHuSOFAT-stimulated osteoclast cultures. IL-6 neutralization failed to have an effect on osteoclast formation induced by rhSOFAT (Figures 6C and D). We further stimulated RAW 264.7 cells with rHuSOFAT for 24 hours and quantitated IL-6 production using a commercial mouse IL-6–specific ELISA. The results showed that, while lipopolysaccharide potently induced IL-6 secretion, rHuSOFAT had no effect (details available from the corresponding author upon request), suggesting that SOFAT does not mediate osteoclast formation through autologous production of IL-6 by monocytes.

Finally, RANKL-induced production of c-fos induces nuclear factor of activated T cells c1 (NF-ATc1) synthesis, which has been described as a master switch for osteoclastogenesis (38). As with RANKL, SOFAT-induced osteoclast formation was sensitive to CSA and FK-506, which are 2 potent inhibitors of calcineurin, the major upstream regulator of NF-ATc1 activation (details available from the corresponding author upon request). SOFAT signaling, like that of RANKL, thus appears to converge on the NF-AT signal transduction pathway. This is consistent with the induction of up-regulated expression of TRAP, β3 integrin, and cathepsin K by SOFAT, all of which are factors reported to have NF-ATc1 consensus sequences in their gene promoters (39).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Threonine synthase is a pyridoxal-5′-phosphate–dependent enzyme that synthesizes L-threonine in plants and microorganisms. However, L-threonine is an essential amino acid, since it is not biosynthetically generated in insects, birds, and mammals. Interestingly, 2 homologs of threonine synthase, THNSL1 and THNSL2, have nonetheless been conserved throughout evolution and persist in rodents and mammals, including humans (33). The high degree of sequence conservation of threonine synthase genes through evolution is surprising, given the apparent lack of L-threonine synthesis in higher organisms. The full-length THNSL2 protein has been reported to bind pyridoxal-5′-phosphate and O-phospho-homoserine, degrading them to alpha-ketobutyrate, phosphate, and ammonia, and degrading O-phospho-threonine to alpha-ketobutyrate, suggesting that the protein exhibits catabolic phospholyase activity (33). Interestingly, SOFAT does not contain the entire pyridoxal-5′-phosphate binding domain, suggesting that unlike threonine synthase, SOFAT is not pyroxidal-5′-phosphate dependent.

Our results suggest that a secreted fragment coded by the threonine synthase homolog may play a critical, previously unrecognized function in the immune/ skeletal interface. Furthermore, the enhanced expression of this cytokine may drive inflammation and bone destruction in pathologic conditions that are characterized by exuberant T cell activation, such as in RA.

We have previously reported that activated T cells produce factors capable of inducing osteoblastic IL-6 production (14) as well as factors that stimulate osteoclastogenesis, including RANKL (27, 28). However the role of RANKL in inflammatory bone destruction is not clear. RANKL has been implicated in systemic bone destruction in animal models of RA (26). However, patients with RA have elevated levels of serum OPG (40), and the OPG:RANKL ratio tends to be high in the synovium. Furthermore, there is often no correlation between the levels of synovial RANKL and disease severity (41). It is thus possible that SOFAT is a critical osteoclastogenic factor in RA.

TNFα has also been found to induce osteoclasts synergistically with RANKL (42) as well as to act independent of RANKL (43), although these latter effects are controversial. Although SOFAT was not found to require TNFα to induce osteoclast formation, the presence of TNFα potently amplified osteoclastogenesis. We speculate that the high levels of TNFα present in RA may act to potently magnify SOFAT-induced osteoclastogenesis and bone loss.

In T cells, the RANKL gene is regulated by a calcineurin-dependent signaling pathway (44). In contrast, SOFAT is secreted by activated T cells via a calcineurin-independent pathway, as demonstrated by findings indicating that CSA treatment of activated T cells failed to block production of this cytokine (14). Taken together, these observations suggest that SOFAT production by T cells is stimulated by an intracellular pathway different from that utilized by RANKL, although the specific pathways involved remain to be elucidated.

The role of IL-6 in bone resorption has been primarily associated with postmenopausal osteoporosis, but IL-6 may also play a role in the bone loss seen in RA and other inflammatory arthritides (20). Our results demonstrated that SOFAT is unable to stimulate, or augment, osteoclast formation by directly inducing IL-6 secretion by osteoclast precursors. However, IL-6 has been reported to stimulate RANKL production by synovial fibroblasts derived from patients with RA and to mediate TNFα-induced RANKL production in this system (45). IL-6 also functions to induce the proliferation of mononuclear osteoclast precursor cells (46). The demonstration that SOFAT is a potent inducer of IL-6 by osteoblasts suggests that SOFAT could play a significant role in the local inflammatory response, and also could exacerbate bone destruction in RA indirectly through multiple IL-6–mediated events. Interestingly, results of a new study suggest that IL-6 modulates production of T cell–derived cytokines in antigen-induced arthritis and drives inflammation-induced osteoclastogenesis (47). These findings suggest the potential existence of a feedback mechanism between activated T cells and osteoblasts, and SOFAT could play a critical role in the coupling of bone cells to the adaptive immune response, thus perpetuating inflammation and osteoclastic bone resorption.

SOFAT may represent the first in a potential family of novel cytokines possessing biologic activities in the absence of classic cytokine-like motifs. The pathophysiologic functions and mechanisms of action of SOFAT remain to be elucidated.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

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. Rifas, Weitzmann.

Acquisition of data. Rifas, Weitzmann.

Analysis and interpretation of data. Rifas, Weitzmann.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

We thank Ms Theresa Geurs for technical assistance, Dr. Henry Rohrs for technical assistance with the mass spectrometry, and Dr. Philip Osdoby for critical reading of the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
ART_24877_sm_SupplFigs.doc232KSupplemental Figures 1-2

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.