This protocol was approved by the Institutional Review Board of Seoul National University Hospital (#1406-043-584). Human peripheral blood and synovial fluid were drawn from healthy volunteers (24) and patients with RA after obtaining written informed consent in accordance with the Declaration of Helsinki.

Peripheral blood mononuclear cells in heparinized peripheral blood and SFMCs in heparinized joint fluid were purified using a Ficoll-Histopaque gradient (1.077 g/mL; GE Healthcare Bio-Sciences, Piscataway, NJ, USA). CD14⁺ monocytes were enriched from SFMCs, which was possible due to the monocytes’ ability to stick to the surface of tissue culture dishes. Briefly, SFMCs were plated on culture dishes for 1 h, then detached using 0.02% EDTA in cold phosphate-buffered saline. The purification of CD14⁺ monocytes was >95%, as confirmed by flow cytometry. PBMCs, SFMCs, and purified CD14⁺ monocytes were grown in α-minimum essential media (MEM) containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (i.e., α-MEM complete medium, all from Life Technologies, Carlsbad, CA, USA).

RAW264.7 cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and grown in Dulbecco’s modified Eagle’s medium (DMEM, Life Technologies) containing 10% FBS and 1% penicillin/streptomycin (i.e., DMEM complete medium). For osteoclast differentiation, RAW264.7 cells were cultured in α-MEM complete medium.

The following antibodies (Abs) were used for flow cytometry staining: phycoerythrin–anti-CD14 (BD Biosciences, San Jose, CA, USA), allophycocyanin-anti-IL-7Rα (eBioscience, San Diego, CA, USA), Alexa Fluor 700–anti-CD3 (eBioscience), and fluorescein isothiocyanate–anti-phospho-STAT5 (p-STAT5) (BD Biosciences). To analyze IL-7Rα expression, cells were stained with Abs to CD3, CD14, IL-7Rα or isotype control. To measure intracellular p-STAT5, cells were first stained with Abs to surface antigens and fixable viability dye eFluor 506 (eBioscience) to irreversibly label dead cells, then stimulated for 30 min with recombinant human IL-7 (10 ng/mL, PeproTech, Rocky Hill, NJ, USA) or PBS. The cells were then fixed with 2% paraformaldehyde for 10 min and permeabilized with 90% methanol for 30 min. Subsequently, the cells were stained with an Ab to p-STAT5. The stained cells were analyzed on a BD™ LSRII (BD Biosciences) with the FACSDiva™ software. After gating on CD3⁺ or CD14⁺ cells from live cells, expression levels of IL-7Rα and p-STAT5 were analyzed in T cells and monocytes.

Interleukin-7Rα-overexpressing RAW264.7 cells (i.e., RAW-IL7ROE cells) were established using lentiviral vectors (Clontech Laboratories; Mountain View, CA, USA). cDNA for IL7RA was amplified from a cDNA clone using PCR (see Table S1 in Supplementary Material for primer sequences). The PCR product was cloned into the pLVX-IRES-ZsGreen1 vector, the destination lentiviral vector for the overexpression of IL-7Rα.

Lentiviruses were produced by co-transfection of 293FT cells (Life Technologies) with the cloned lentiviral vector, together with three packaging plasmids (pLP1, pLP2, pLP/VSVG) using PromoFectin (Promokine, Heidelberg, Germany). Then, the viruses were collected and concentrated using PEG-it (System Biosciences, Palo Alto, CA, USA). Cells were infected with the cloned lentivirus in the presence of 8 µg/mL hexadimethrine bromide (Life Technologies) by centrifugation (1,200 × g, 1 h, 32°C). To collect lentiviral-transduced cells, cells expressing a GFP reporter were sorted using a BD FACSAria system (BD Biosciences).

Cells were cultured in 384-well culture plates in the presence of recombinant human macrophage colony-stimulating factor (M-CSF) (20 ng/mL), recombinant human RANKL (50 ng/mL), or IL-7 (2 ng/mL; all from PeproTech) for the indicated number of days by replacing the medium more than 90% of the original medium at 3-day intervals with fresh cytokines, thereby removing the effect of previously produced cytokines and unattached cells, including T and B cells to leave only adherent osteoclast precursor cells. Otherwise, cells were cultured with IL-7 (2 ng/mL) for 3 days, followed by treatment with M-CSF (20 ng/mL), RANKL (50 ng/mL), or IL-7 (2 ng/mL) by replacing the medium, as described above, to evaluate the effect of IL-7 pretreatment. OPG was purchased from PeproTech to inhibit RANKL/RANK signaling. The JAK inhibitor, tofacitinib, and a STAT5 inhibitor (CAS 285986-31-4; Sigma-Aldrich, St. Louis, MO, USA) were used to evaluate JAK/STAT signaling.

After 5–15 days of culture, cells were fixed and stained for tartrate-resistant acid phosphatase (TRAP) using an acid phosphatase, leukocyte kit (Sigma-Aldrich) according to the manufacturer’s protocol. Mature osteoclasts, which were defined as multinucleated (≥3 nuclei) TRAP⁺ cells, were counted manually with a light microscope (Olympus, Tokyo, Japan). Images were captured using the ProgRes Capture Pro software (Jenoptik, Jena, Germany).

Wild-type STAT5a (STAT5A-WT), dominant-negative STAT5a/b (STAT5A-DN, STAT5B-DN), and a constitutively active STAT5a (STAT5A-CA) (25) were transiently transfected into RAW274.7 or RAW-IL7ROE cells using electroporation (Bio-Rad, Hercules, CA, USA). Then, cells were cultured with recombinant mouse RANKL (50 ng/mL) in the presence or absence of recombinant mouse IL-7 (2 ng/mL) for 6 days by changing the medium at 3-day intervals with fresh cytokines. At day 6, cells were stained with TRAP.

The functional resorption activity of the differentiated osteoclasts was evaluated by resorption pit formation, as described below. Cells were cultured for 30 days on top of dentine disks (Immunodiagnostic Systems, Boldon, UK) in 96-well culture plates in the presence of M-CSF (20 ng/mL), RANKL (50 ng/mL), or IL-7 (2 ng/mL) to differentiate osteoclasts. Then, the dentine slices were washed three times and immersed in 70% sodium hypochlorite to remove adherent cells. The resorption lacunae were counterstained with 1% (w/v) toluidine blue in 0.5% sodium borate for 60 s (Sigma-Aldrich). Photographs were taken through an LSM 5 PASCAL confocal microscope (Carl Zeiss, Jena, Germany) to analyze the surface topography and the area of the resorption pits was measured in four randomly selected areas for each dentine slice with the LSM 5 Image Browser (Carl Zeiss). Roughness parameters obtained were:roughness average (Ra), which is the main height calculated over the entire measured length or area, Rq, statistical moments of peak distribution (symmetry), Rz, mean roughness depth, and Rv, maximum profile valley depth, which are the distances from the mean line/surface to the highest/lowest point in the evaluation length/area (26).

For quantitative PCR (qPCR), total RNA was extracted from cells using TRIzol reagent (Life Technologies). Extracted RNA was used for cDNA synthesis using a Transcript First Strand cDNA synthesis Kit (Roche Applied Science, Basel, Switzerland). PCR was performed using qPCR PreMix (Bioneer, Daejeon, Republic of Korea) (see Table S1 in Supplementary Material for primer sequences). qPCR was performed with an Exicycler 96 Quantitative Real-Time PCR System (Bioneer). Differences in expression were normalized to expression of the control gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or ribosomal protein S18 (RPS18).

Protein lysate was resolved by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane. The following Abs were used for immunoblotting: rabbit monoclonal Abs to phospho (p)-Akt (Ser473), Akt, extracellular signal-regulated kinases (Erk), p-Erk (Thr202/Tyr204) (all from Cell Signaling Technology, Danvers, MA, USA), p-STAT5 (EMD Millipore, Billerica, MA, USA), and STAT5 (BD Biosciences). The membranes were stained with appropriate Abs and visualized using SuperSignal West Femto Chemiluminescent Substrate (Thermo Fisher Scientific, Fremont, CA, USA).

All data are expressed as means ± SEM. Data were compared with the control group using a two-tailed Student’s t-test and two-way analysis of variance with Bonferroni post hoc test. p values <0.05 were considered to indicate statistical significance. All statistical analyses were performed using the GraphPad Prism software (ver. 6.01; GraphPad Software, La Jolla, CA, USA).

Although IL-7 is known to stimulate osteoclast formation by augmenting the production of TNF-α and RANKL by T cells (14–16, 27), there has been little research on the characteristics of osteoclasts induced by IL-7 or the characteristics of progenitor cells derived under inflammatory conditions. Thus, we explored how IL-7 affected osteoclast formation by analyzing the pattern of multinucleated TRAP⁺ cells (≥3 nuclei), and measuring roughness using various parameters for dentine pits. PBMCs from healthy individuals and SFMCs from patients with RA were cultured for 15 or 30 days with M-CSF, RANKL, and IL-7 to determine osteoclast differentiation. We found that osteoclasts from PBMCs whose differentiation was induced by M-CSF and RANKL were typically giant multinucleated TRAP⁺ cells by day 10 (Figures 1A,B). However, IL-7-mediated TRAP⁺ cells were small multinucleated cells that appeared even earlier, on day 5 (Figure 1A), a differentiation that was similarly observed in SFMCs (Figure S1A in Supplementary Material). On day 10, however, osteoclast differentiation from PBMCs by IL-7 was comparable to that from M-CSF and RANKL (Figure 1B), while osteoclast formation from SFMCs by IL-7, although meaningful compared to the M-CSF control, was significantly lower than that induced by M-CSF and RANKL (Figure S1B in Supplementary Material). Overall, osteoclast formation from SFMCs was superior to PBMCs in terms of the number of TRAP⁺ cells. Although the size of the multinucleated TRAP⁺ cells by IL-7 was small, we found that the roughness of dentine pit formation by IL-7-induced osteoclasts was comparable with the pit formation by M-CSF- and RANKL-induced osteoclasts, regardless of whether the osteoclasts were derived from PBMCs or SFMCs (Figures 1C,D; Figures S1C,D in Supplementary Material). That is, when considering roughness parameters—Ra, Rq, Rz, and Rv—the depth of the pits formed by the IL-7, M-CSF, and RANKL-induced osteoclasts were equivalent (Figure 1D; Figure S1D in Supplementary Material) although the area of the pits, as shown by toluidine blue staining, was larger for M-CSF and RANKL than for IL-7 (Figure 1C; Figure S1C in Supplementary Material). Next, we further analyzed the expression of osteoclast markers [cathepsin K (CathK) and RANK] induced by IL-7, M-CSF, and RANKL. IL-7 induced the expression of CathK and RANK comparable to M–CSF and RANKL (Figure 1E; Figure S1E in Supplementary Material), suggesting that IL-7 can induce osteoclast differentiation through a mechanism similar to RANKL.

We have demonstrated that IL-7-mediated osteoclasts are unique in TRAP⁺ cell size and functional resorption, such as dentine pit formation. These findings suggest that IL-7-induced osteoclast formation might occur by a mechanism different from the previously known mechanisms involving the induction of TNF-α and RANKL by T cells (14–16).

These results raised a question about whether the fate of IL-7-induced osteoclasts could be altered by the osteoclastogenic factor, RANKL, because osteoclasts formed by IL-7 differed from those by RANKL in many ways (Figure 1; Figure S1 in Supplementary Material). To address this, we pretreated cells with IL-7, then treated them with M-CSF, RANKL, or IL-7, replacing more than 90% of the original medium at 3-day intervals to remove the effect of previously produced cytokines and unattached cells. Through preliminary experiments, we found that most T and B cells were removed by the third day of media replacement (data not shown), which suggested that in our experimental conditions, the removal would take 3 days even if IL-7 acts on T cells, as reported previously (14). Surprisingly, the small, multinucleated TRAP⁺ cells of IL-7-induced osteoclasts (Figure 1; Figure S1 in Supplementary Material) were not altered when RANKL was added after IL-7 pretreatment (Figure 2A). However, IL-7 pretreatment induced osteoclast formation to a level similar to that of RANKL treatment (Figure 2B), and the fully differentiated osteoclasts formed dentine pits (Figure 2C). Interestingly, after IL-7 pretreatment, M-CSF alone resulted in the formation of TRAP⁺ cells, which were capable of forming dentine pits (Figures 2A,C). However, continuous treatment with IL-7 led to the formation of dentine pits deeper than those formed by cells treated with RANKL following IL-7 pretreatment (Figure 2C). This raised the possibility that IL-7 may act directly and uniquely on osteoclasts precursors, a notion that would be supported if IL-7Rα was expressed on osteoclast precursors.

Based on our observations and previous reports (22, 23), we proposed the idea that IL-7Rα could be induced by inflammatory factors produced by T or B cells even if IL-7Rα is not expressed in progenitor cells. As expected, we found that the frequency of IL-7Rα⁺ CD14⁺ monocytes was increased significantly in the synovial fluid of RA patients compared with peripheral blood from both RA patients and healthy individuals, where IL-7Rα was rarely expressed (Figure 3A). Of particular importance, this increased IL-7Rα on monocytes reacted with IL-7 to phosphorylate STAT5 (Figure 3B), a finding that raised the possibility of direct osteoclast differentiation by IL-7 for precursor cells that express IL-7Rα. In particular, stimulation of PBMCs from healthy individuals with IL-7 resulted in increased expression of IL-7Rα, as well as STAT5 activation in monocytes (Figure S2 in Supplementary Material). This suggests that IL-7 acts on IL-7Rα-expressing cells first, thereby inducing functional IL-7Rα expression in CD14⁺ cells and ultimately stimulating osteoclast formation.

Next, to confirm whether IL-7 directly induced TRAP⁺ cells of IL-7Rα⁺ monocytes, we isolated CD14⁺ monocytes expressing IL-7Rα from SFMCs of RA patients and cultured them with IL-7. Remarkably, these CD14⁺ monocytes differentiated into TRAP⁺ cells with IL-7 (Figure 3C); the degree of osteoclast differentiation was similar to that with M-CSF and RANKL (Figure 3C). This finding demonstrates for the first time that IL-7 alone can induce sufficient osteoclast differentiation from human monocytes that express IL-7Rα.

Interleukin (IL)-7 has been shown to upregulate RANKL secretion in T cells (14), which led us to ask whether IL-7-induced osteoclast formation was due to the presence of secreted RANKL. To exclude this possibility, we added the RANKL inhibitor OPG, a naturally occurring secreted protein (28), into cultures of PBMCs and SFMCs treated with IL-7. We found that OPG completely inhibited RANKL-induced osteoclast formation (Figures 3D,E), but hardly blocked osteoclast formation from PBMCs by IL-7 (Figure 3D). However, OPG repressed IL-7-induced osteoclast formation from SFMCs by approximately 60% (Figure 3E; mean ± SEM = 75.2 ± 28.8 vs. 29.8 ± 6.2). Thus, OPG did not inhibit IL-7-induced osteoclast formation, although it partially blocked osteoclast differentiation from SFMCs, suggesting that IL-7 can directly induce osteoclast formation in a RANKL-independent manner.

Based on our observations, we investigated the mechanism by which IL-7 induced osteoclast differentiation via IL-7Rα using RAW264.7 cells, a mouse leukemic monocyte-macrophage cell line. We constructed IL-7Rα-overexpressing RAW264.7 cells (RAW-IL7ROE) using lentiviral transduction (Figure 4A) and determined whether they differentiated into osteoclasts in response to IL-7. We found that IL-7 induced osteoclast differentiation of RAW-IL7ROE cells, but not control (RAW-vector) cells (Figure 4B). However, IL-7, in combination with RANKL, induced multinucleated TRAP⁺ cells in both RAW-IL7ROE and RAW-vector cells in comparable numbers (Figure 4B). IL-7 in combination with RANKL significantly increased the number of dentine pits formed by osteoclasts differentiated from RAW-IL7ROE cells compared with IL-7 alone (Figure 4C), suggesting that IL-7 could induce osteoclast formation by a different mechanism than RANKL. We also found that OPG completely inhibited RANK-induced TRAP⁺ cells, but did not affect the formation of TRAP⁺ cells by IL-7 or IL-7 in combination with RANKL (Figure S4 in Supplementary Material). Thus, these findings confirmed that IL-7 alone can cause differentiation into fully functioning osteoclasts.

To address the underlying mechanisms, we analyzed the expression of osteoclast markers [TRAP, CathK, tyrosine-protein kinase SRC-1 (c-src), and calcitonin receptor (CalcR)] induced by IL-7 and RANKL. Overall, IL-7 induced the expression of osteoclast marker genes at similar levels to RANKL (Figures S3A,B in Supplementary Material). Furthermore, we explored whether the expression of nuclear factor of activated T-cells 1 (NFATc1), a master regulator gene that is upregulated by RANKL, was affected by IL-7 during osteoclastogenesis because RANKL upregulates NFATc1 via the expression of the early osteoclastic gene, c-Fos, which, in turn, induces osteoclast marker genes (29). We found that IL-7 induced expression of NFATc1 mRNA at a level comparable to that of RANKL, but there was no synergistic effect with RANKL (Figure 4D). Together, this suggests that IL-7 can induce the differentiation of functional osteoclasts similar to that of the osteoclastogenic factor RANKL.

Finally, to explore how IL-7 induces osteoclast differentiation through IL-7Rα in progenitors in more detail, we analyzed components of early IL-7R signaling, including p-STAT5, p-Akt, and p-Erk, to determine which signaling molecules were uniquely activated by IL-7, and not by RANKL. We discovered that activation of Akt and Erk was common to RANKL and IL-7, whereas activation of STAT5 was unique to IL-7 (Figures 5A,B), suggesting that STAT5 could be a key molecule in IL-7-induced osteoclast formation.

Based on these observations, we transfected constitutively active STAT5a (STAT5A-CA) or dominant-negative STAT5a/b (STAT5A-DN or STAT5B-DN) plasmids into RAW-vector and RAW-IL7ROE cells to further analyze the role of STAT5 in IL-7-induced osteoclast formation. When RAW-IL7ROE cells were transfected with STAT5A-DN or STAT5B-DN, RANKL, or IL-7 in combination with RANKL-induced osteoclast formation, but IL-7 did not (Figures 5C,D). These findings indicate that STAT5 is a key molecule in IL-7-mediated osteoclast formation. Interestingly, while STAT5A-CA transfection alone was able to induce differentiation of multinucleated TRAP⁺ cells in the absence of stimulation (Figure 5C), it failed to induce dentine pit formation and even repressed dentine pit formation by RANKL (Figure 5D). These data suggest that despite the fact that the activation of STAT5 by IL-7 can produce functional osteoclasts that form dentine pits, inappropriate or excessive STAT5 activity in the absence of IL-7 may inhibit RANKL-induced osteoclast formation.

Finally, we explored the clinical utility of IL-7R signaling in IL-7-mediated osteoclast formation. To address this, we used the JAK inhibitor tofacitinib, which was the first drug found to interfere with the JAK-STAT signaling pathway (30), and a STAT5 inhibitor, which interferes with DNA binding of STAT5 (31). We found that tofacitinib markedly inhibited IL-7-induced osteoclast formation from PBMCs as well as SFMCs, but not by RANKL (Figures 6A,B). The STAT5 inhibitor also considerably suppressed osteoclast formation in response to IL-7 in both PBMCs and SFMCs (Figures 6C,D). Furthermore, IL-7-induced osteoclast formation through the JAK/STAT signaling pathway was again confirmed by inhibitor treatment in RAW-IL7ROE cells (Figure S4 in Supplementary Material). Together, these data suggest that the IL-7R signaling pathway may serve as a target for the treatment of bone erosion by IL-7.