SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Objective

To examine the suppressive effect of anti-human Fas monoclonal antibody (mAb) on osteoclastogenesis in rheumatoid arthritis (RA) both in vitro and in vivo.

Methods

For in vitro analysis, activated CD4+ T cells derived from peripheral blood mononuclear cells were left untreated or were treated with humanized anti-human Fas mAb (R-125224) and cocultured with human monocytes. On day 12, the number of tartrate-resistant acid phosphatase (TRAP)–positive multinucleated cells was counted. For in vivo analysis, tissue derived from human RA pannus was implanted with a slice of dentin subcutaneously in the backs of SCID mice (SCID-HuRAg-pit model). R-125224 was administered intravenously once a week for 3 weeks. The implanted tissue and dentin slice were removed, and the pits formed on the dentin slice were analyzed.

Results

In vitro, coculture of activated CD4+ T cells and peripheral monocytes induced osteoclastogenesis. The number of TRAP-positive multinucleated cells was reduced when activated CD4+ T cells were treated with R-125224. We established a new animal model for monitoring osteoclastogenesis, SCID-HuRAg-pit. We found that with R-125224 treatment, the number of pits formed on the implanted dentin slices was significantly reduced and the number of lymphocytes in the implanted RA synovial tissue was dramatically reduced in this model.

Conclusion

This is the first study to demonstrate the suppressive effect of anti-human Fas mAb on osteoclastogenesis in RA synovial tissues through the induction of T cell apoptosis. Induction of apoptosis of infiltrated lymphocytes could be a useful therapeutic strategy for RA, in terms of suppressing both inflammation and bone destruction.

Rheumatoid arthritis (RA) is characterized as both an autoimmune reaction initiated by lymphocytes and a proliferation of inflamed synovial membrane accompanied by inflammatory cell infiltration and bone destruction (1, 2). Bone resorption at the site of lesions in RA is caused by osteoclasts, and this site is the so-called bare area (3). Osteoclasts are multinucleated cells that are formed by the fusion of mononuclear cell precursors (4). Osteoclast precursors, which express the receptor activator of NF-κB (RANK) (5–7), recognize the RANK ligand (RANKL) (8) through cell-to-cell interactions with osteoblasts and/or stromal cells, and they differentiate into osteoclasts in the presence of macrophage colony-stimulating factor (M-CSF) (9, 10). RANKL is also expressed on activated T cells, and activated T cells directly induce osteoclast differentiation in vitro (11, 12). Furthermore, a T cell–derived cytokine, interleukin-17 (IL-17), the levels of which are elevated in the synovial fluid of RA patients, is reported to induce osteoclast differentiation (13). These results indicate that in autoimmune arthritis such as RA, bone resorption is regulated by the immune system (14, 15).

The Fas antigen (CD95), a cell-surface receptor that belongs to the tumor necrosis factor receptor/nerve growth factor receptor family, transduces a cell death signal (16). Fas antigen and Fas ligand have been considered to play an important role in the development of autoimmune diseases because a mutation in Fas and Fas ligand leads to immunologic disorders associated with lymphadenopathy and proliferative arthritis in mice and humans (17–20). Fas is expressed on activated lymphocytes and various other cells, including rheumatoid synovial cells (21). Furthermore, stimulation of Fas by Fas ligand or agonistic anti-Fas monoclonal antibody (mAb) induces apoptosis in various cells (21–25). Thus, it is possible that agonistic anti-Fas mAb may have therapeutic effects in RA. In fact, the anti-mouse Fas antibody RK-8 was shown to effectively ameliorate collagen-induced arthritis in the mouse (26). The anti-human Fas antibody CH-11 was also shown to be effective in SCID mice that had been engrafted with human RA tissues (SCID-HuRAg) (27).

One of the serious problems linked with anti-Fas mAb is hepatotoxicity. We have previously reported that our novel humanized anti-human Fas mAb R-125224, which originated from m-HFE7A (28), induced apoptosis in activated lymphocytes but not in hepatocytes. Administration of R-125224 to SCID-HuRAg mice was shown to reduce the number of inflammatory cells (29). Thus, R-125224 could be a candidate therapeutic agent for use in RA.

In the present study, we investigated whether the induction of apoptosis in activated T cells by R-125224 could suppress osteoclastogenesis in vitro and in vivo. Since there are no useful in vivo animal models for monitoring osteoclastogenesis, we established a novel animal model by modifying the SCID-HuRAg model. Dentin slices were engrafted with RA synovial tissues subcutaneously on the back of SCID mice (SCID-HuRAg-pit). Pits formed on the dentin slices 3 weeks after transplantation. Administration of R-125224 to SCID-HuRAg-pit reduced the number of infiltrated lymphocytes in the transplanted tissues and prevented the formation of pits on the dentin slices. These findings suggest that induction of apoptosis in the infiltrated lymphocytes could be a useful therapeutic strategy for RA, in terms of suppressing both inflammation and bone destruction.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Animals and antibodies.

Male CB17/Icr Crj-SCID (SCID) mice ages 4–6 weeks were purchased from Charles River Japan (Yokohama, Japan). The mice were housed in specific pathogen–free facilities, and water and food were provided ad libitum.

R-125224 is a humanized anti-human Fas antibody derived from m-HFE7A (27). Human IgG was purchased from Cosmo Bio (Tokyo, Japan).

Flow cytometry analysis of osteoclasts and activated CD4+ T cells.

Human peripheral blood from healthy volunteers was collected into syringes containing preservative-free heparin. Peripheral blood mononuclear cells (PBMCs) were isolated by centrifugation over Ficoll-Paque Plus (Amersham Biosciences, Uppsala, Sweden) density gradients, washed, and resuspended at 2 × 106 cells/ml. PBMCs were cultured for 2 hours, and nonadherent cells were used for separation of the CD4+ T cells. CD4+ T cells were separated by magnetic sorting. The separated CD4+ T cells were activated with phytohemagglutinin (5 μg/ml; Sigma, St. Louis, MO) and IL-2 (10 ng/ml; Genzyme, Cambridge, MA) in 6-well plates for 1 day. For flow cytometry analysis, activated CD4+ T cells were incubated with fluorescein isothiocyanate (FITC)–conjugated anti-human CD4 antibody (PharMingen, San Diego, CA) or FITC-conjugated anti-human Fas antibody for 15 minutes at 4°C.

Osteoclasts were differentiated from the adherent cells obtained from PBMCs as described above, by culturing for 10 days in flasks in the presence of M-CSF (25 ng/ml), soluble RANKL (100 ng/ml), dexamethasone (10–8M), and transforming growth factor β (10 ng/ml). Osteoclasts were collected and incubated with FITC-conjugated anti-human CD51/61 antibody (PharMingen) or FITC-conjugated anti-human Fas antibody (MBL, Nagoya, Japan) for 15 minutes at 4°C.

Stained activated CD4+ T cells and osteoclasts were washed and analyzed using a FACScan (Becton Dickinson, Mountain View, CA). For the apoptosis-induction assay, the activated CD4+ T cells were incubated with and without R-125224 (1,000 ng/ml) for 2 hours, washed, and resuspended in RPMI 1640 supplemented with 10% fetal bovine serum in the presence of anti-human IgG (500 ng/ml) for crosslinking, and incubated for 18 hours. At the end of the incubation, XTT (2,3-bis-[2-methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxanilide; final concentration 0.2 mg/ml) and phenazine methosulfate (final concentration 5 μM) were added to determine cell viability (28).

Coculture system for osteoclastogenesis and R-125224 treatment.

Human PBMCs were obtained as described above and cultured for 2 hours. Nonadherent cells were used for separation of CD4+ T cells. Adherent cells were collected and cultured for 3 days in 48-well plates (1 × 105 cells/0.3 ml/well) in the presence of M-CSF (25 ng/ml; Genzyme) before activated CD4+ T cells were added to the wells.

CD4+ T cells were separated from the nonadherent cells by magnetic sorting. The separated CD4+ T cells were cultured for 2 days and activated with phytohemagglutinin (5 μg/ml) and IL-2 (10 ng/ml) in 6-well plates for 1 day. The activated CD4+ T cells were incubated with and without R-125224 (1,000 ng/ml) for 2 hours, washed, and resuspended in RPMI 1640 supplemented with 10% fetal bovine serum in the presence of anti-human IgG (500 ng/ml) for crosslinking. Activated CD4+ T cells treated with and without R-125224 were added at 1 × 104 cells/well to the 3-day culture of adherent cells and cocultured in the presence of 1,000 ng/ml of anti–interferon-γ (anti-IFNγ) mAb (Genzyme) for 12 days. Culture samples were incubated in triplicate. Osteoclast formation was evaluated by tartrate-resistant acid phosphatase (TRAP) staining, and the number of TRAP-positive multinucleated cells that had more than 4 nuclei was counted.

Preparation of SCID-HuRAg-pit mice, a novel animal model for the evaluation of osteoclastogenesis.

We established a new animal model for monitoring osteoclastogenesis in vivo by modifying the SCID-HuRAg model, which has been shown to closely reflect the features of RA in humans (29, 30). Synovial tissues obtained during synovectomy or total joint replacement surgery in RA patients were used for implantation. Informed consent was obtained from all patients. Before implantation, the inflammatory activity in the synovial tissues was examined by hematoxylin and eosin staining, and lymphocyte-rich synovium was used for the experiment.

A slice of dentin (Kureha Special Laboratory, Iwaki, Japan) was covered with the RA synovium, and the tissues were grafted subcutaneously on the back of SCID mice that had been anesthetized by inhalation of ethyl ether. We named this the SCID-HuRAg-pit mouse model. The weight of the synovium sample was adjusted to 0.5 gm (±0.05 gm) beforehand. All surgical procedures were performed under sterile conditions.

Mice were euthanized 2, 3, and 4 weeks after implantation, and the implanted tissues and dentin slices were removed. The dentin slices were washed, and resorption pits were stained with acid hematoxylin. The resorption pits and osteoclasts on the dentin slices were examined by electron microscopy.

R-125224 treatment protocol.

We used synovial tissues obtained from 3 different RA patients. The dentin slices were covered with the synovial tissue samples and transplanted into the SCID mice. The synovial tissue from each patient was used to provide synovial–dentin samples for 2–6 SCID mice. A total of 13 SCID mice were used for the R-125224 treatment study. The SCID-HuRAg-pit mice were then randomly divided into 2 groups. One group was treated intravenously once a week for 3 weeks after implantation (n = 6) with 10 mg/kg of R-125224; the other group was given 10 mg/kg of human IgG as a control (n = 7). The mice were euthanized 1 week after the last injection, and the implanted tissues and dentin slices were removed.

The number of 0.1-mm2 areas of a grid that were fully or partially covered by the resorption pits formed on the dentin slice was counted visually under a microscope. The synovial tissues were fixed for 2 hours in 4% paraformaldehyde, cut into sections, embedded in paraffin, and mounted on slides. The slides were deparaffinized in xylene and rehydrated through a series of alcohol solutions. Slides were treated with 0.3% aqueous H2O2 for 10 minutes to quench the endogenous peroxidase, boiled in citric acid buffer to remove excess antigen, and blocked with 10% normal rabbit serum for 15 minutes. Then, the slides were incubated overnight at 4°C with a mAb against human CD4 (Nichirei, Tokyo, Japan). After overnight incubation, slides were incubated with biotinylated secondary antibody (Nichirei) for 30 minutes and then with peroxidase-labeled streptavidin (Nichirei) for 30 minutes. After these treatments, the slides were stained for 5 minutes with diaminobenzidine dissolved in 50 mM Tris buffer (pH 7.5). The number of CD4+ cells in a total of 20 focal areas was counted with the use of a microscope (at 400× magnification).

Statistical analysis.

Statistical analysis of the number of pits was performed by the randomization test. Since 3 different RA patients provided synovial tissue samples for the control and R-125224 treatment groups, the data from the 3 experimental units (1 for each patient) were examined.

The randomization test was conducted as follows. First, the mean number of pits in the R-125224 treatment group was subtracted from that of the control group for each experimental unit, and the sum of these differences in the means (Tobs) was calculated.

Next, while maintaining the same sample size in the control and treatment groups and the same number of experimental units as that for the observed values, all possible combinations (400 allocations in this case) of data were considered. For each combination, the sum of the differences in the means for the 3 experimental units (Ti, where i = 1–400) was calculated as described above.

Then, the number of Ti values that were greater than or equal to the Tobs value was counted (defined as n). Finally, dividing n by the number of possible allocations (i.e., 400 allocations) gave the P value for this randomization test. P values less than 0.05 indicated that the Tobs value was not obtained by chance and that the efficacy of R-125224 was significant.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Fas expression on osteoclasts and activated CD4+ T cells.

We have previously reported that the humanized anti-human Fas antibody R-125224 induced apoptosis in Fas-expressing activated human PBMCs with crosslinking (28), but it has not been determined whether osteoclasts expressed Fas and whether R-125224 induced apoptosis in them. Thus, we used a flow cytometry method to examine whether osteoclasts expressed Fas. Osteoclasts were differentiated in vitro as described in Materials and Methods. Twelve days later, TRAP-positive multinucleated cells formed, and these cells were found to express the vitronectin receptor (CD51/61), one of the markers for osteoclasts. CD51/61+ osteoclasts did not express Fas (Figure 1A). Activated CD4+ T cells, however, expressed Fas (Figure 1B). These results indicate that osteoclasts were not susceptible to Fas-mediated apoptosis.

thumbnail image

Figure 1. Fas expression of osteoclasts and activated CD4+ T cells. A, Osteoclasts were differentiated in vitro as described in Materials and Methods and stained with fluorescein isothiocyanate (FITC)–labeled control IgG (broken line), FITC-labeled CD51/61 (solid line in left panel), or FITC-labeled Fas (solid line in right panel). B, CD4+ T cells obtained from peripheral blood mononuclear cells were activated with phytohemagglutinin and interleukin-2 for 1 day and then stained with FITC-labeled control IgG (broken lines), FITC-labeled CD4 (solid line in left panel), or FITC-labeled Fas (solid line in right panel).

Download figure to PowerPoint

Suppressive effect of R-125224 on osteoclastogenesis.

Since activated T cells have previously been shown to express RANKL and to directly induce osteoclastogenesis (11, 12), we investigated the ability of R-125224 to suppress osteoclastogenesis through the induction of apoptosis in activated T cells. Adherent PBMCs from healthy volunteers were used as a source of osteoclast precursor cells and were treated with M-CSF to promote osteoclast formation. Adherent PBMCs were cocultured with activated CD4+ T cells from the same volunteers, and 12 days later, TRAP-positive multinucleated cells formed. In this coculture system, the osteoclast formation induced by activated CD4+ T cells was inhibited when activated CD4+ T cells were treated with crosslinked R-125224 (Figure 2A). We also confirmed that when crosslinked, R-125224 showed killing activity of activated CD4+ T cells (Figure 2B).

thumbnail image

Figure 2. A, Suppressive effect of R-125224 on osteoclast formation in cocultures of human adherent cells derived from peripheral blood mononuclear cells and activated CD4+ T cells. After a 12-day culture, tartrate-resistant acid phosphatase–positive osteoclasts were counted. Anti-human IgG (ahIgG) was added for crosslinking. B, Apoptosis-inducing effects of R-125224 on activated CD4+ T cells. Activated CD4+ T cells were treated with and without R-125224 as described in Materials and Methods. Cell viability was determined by XTT assay. Values are the mean and SD of triplicate cultures. Experiments were repeated 4 times with similar results.

Download figure to PowerPoint

Establishment of a novel animal model for the evaluation of osteoclastogenesis in RA, the SCID-HuRAg-pit mouse.

Since there were no useful in vivo models for the evaluation of osteoclastogenesis, we established a new animal model with which to monitor osteoclastogenesis by modifying the existing SCID-HuRAg mouse (29, 30). When we transplanted synovial tissues together with the bones from the same patients, the bones were enucleated and became necrotic, probably due to insufficient vascularization around the implanted bone (data not shown). We therefore decided to use dentin slices, which are commonly used in pit-formation assays in vitro, instead. A dentin slice was covered with synovial tissue from RA patients and transplanted into SCID mice to produce SCID-HuRAg-pit mice. The mice were euthanized 2, 3, and 4 weeks after implantation, and the pit formation on the implanted dentin slices was examined.

As shown in Figure 3, pits on the implanted dentin slices started to form during the third week, and the number of pits was increased at 4 weeks after transplantation. The osteoclasts and pits on the implanted dentin slices were examined by electron microscopy (Figure 4). When a dentin slice was implanted into a SCID mouse without the accompanying RA synovium, no pits were formed at 3 weeks after implantation (data not shown). From these results, the SCID-HuRAg-pit model was confirmed to be a useful animal model for the evaluation of osteoclastogenesis.

thumbnail image

Figure 3. Time course of pit formation on dentin slices in a SCID-HuRAg-pit mouse model of osteoclastogenesis. Dentin slices were removed from the SCID-HuRAg-pit mice 2, 3, and 4 weeks after implantation. Slices were stained with acid hematoxylin, and the resorption pits were identified by microscopy (original magnification × 100).

Download figure to PowerPoint

thumbnail image

Figure 4. Scanning electron micrograph of a dentin slice that had been implanted into the subcutaneous tissue on the back of a SCID-HuRAg-pit mouse, showing the area of resorption that formed on the dentin slice (original magnification × 1,000).

Download figure to PowerPoint

Suppressive effect of R-125224 treatment on osteoclastogenesis in the SCID-HuRAg-pit mouse model.

We examined the suppressive effect of R-125224 on osteoclastogenesis in RA in vivo using a SCID-HuRAg-pit mouse. As shown in Figure 5, hematoxylin and eosin staining of the implanted tissues from the control group treated with human IgG revealed lymphocyte infiltration around vessels. In contrast, the number of infiltrated lymphocytes around the vessels was lower in implanted tissues from the group treated with R-125224. These results were consistent with our previous data (29).

thumbnail image

Figure 5. Reduction in the number of infiltrated lymphocytes by treatment with R-125224. Hematoxylin and eosin staining of implanted tissues from SCID-HuRAg-pit mice treated with A, control IgG and B, R-125224. The number of infiltrated lymphocytes around vessels (arrowhead) was reduced by treatment with R-125224 as compared with the control. (Original magnification × 400.)

Download figure to PowerPoint

Interestingly, a significantly lower number of pits was noted on the implanted dentin slices from the group treated with R-125224 as compared with the number on the implanted dentin slices from the control group treated with human IgG (Figure 6). Similar results were obtained in 3 independent experiments (Figure 7A). The number of pits per slice was statistically significantly different between the two treatment groups (Figure 7B). The number of CD4+ T cells in implanted tissues was also reduced by R-125224 treatment (Figure 7C). This is the first in vivo evidence of anti-human Fas antibody suppression of osteoclastogenesis in RA synovial tissues.

thumbnail image

Figure 6. Reduction in the number of resorption pits formed on dentin slices in SCID-HuRAg-pit mice treated with R-125224. Implanted dentin slices from mice treated with A, control IgG and B, R-125224. The number of resorption pits was reduced by treatment with R-125224 as compared with the control. (Original magnification × 100.)

Download figure to PowerPoint

thumbnail image

Figure 7. Suppressive effect of R-125224 on pit formation in SCID-HuRAg-pit mouse. A, Three lots of human rheumatoid arthritis synovial tissues were transplanted together with dentin slices into SCID-HuRAg-pit mice and used for the R-125224 treatment study. The number of pits formed on the dentin slices was reduced with R-125224 treatment as compared with the control. B, Mean results of the 3 experiments, showing a statistically significant reduction in the number of pits formed on the dentin slices with R-125224 treatment ( = P < 0.05). C, Implanted synovial tissues removed from mice were stained with a monoclonal antibody against human CD4, and the number of CD4+ cells in a total of 20 focal areas was counted microscopically (at 400× magnification). Treatment with R-125224 reduced the number of CD4+ cells.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

In the present study, we used a coculture system to demonstrate that R-125224, a humanized anti-human Fas antibody, inhibited osteoclastogenesis mediated by activated CD4+ T cells. We found that osteoclasts did not express Fas, which indicates that osteoclasts were not the targets of the R-125224–induced apoptosis. CD4+ T cells, on the other hand, expressed Fas and induced osteoclast differentiation from progenitor cells in the coculture system. Thus, the suppression of osteoclastogenesis induced by R-125224 occurred through the induction of apoptosis in infiltrated T cells in the RA synovium.

It was recently reported that activated T cells promote osteoclastogenesis through RANKL expression and regulate bone resorption in autoimmune arthritis (11, 12). Furthermore, other immunomodulatory factors expressed by T cells, such as IL-17, modulate osteoclastogenesis (13). Activated T cells, however, also produce IFNγ, and IFNγ negatively affects osteoclastogenesis (31). Nevertheless, in RA, the production of IFNγ in the synovium is suppressed (32, 33), whereas RANKL expression is enhanced (34, 35). Based on these data, we included anti-IFNγ antibody in our in vitro coculture systems to reflect the clinical situation in RA.

There were no animal models that could be used to evaluate osteoclastogenesis in vivo. We therefore established such an animal model by modifying our SCID-HuRAg mouse (27, 29, 30). The SCID-HuRAg mouse is a model prepared by transplanting human RA synovial tissue into the subcutaneous tissues on the back of SCID mice. Several weeks after the transplantation, neovascularization was observed in the transplanted tissue. The histologic characteristics of the implanted tissue were similar to those of RA in humans, and immunohistologic analysis revealed that cells in the implanted tissue were of human origin (27, 30). Furthermore, potent antirheumatic drugs, such as methotrexate, which are used in clinical settings, were effective in this model (36). These results indicate that the SCID-HuRAg mouse model well reflects the clinical situation.

We had already determined that the cotransplanted bones tended to become necrotic (data not shown), making it difficult to characterize osteoclastogenesis. Pit formation assays on dentin slices are widely used to monitor osteoclast function in vitro, and it has been reported that TRAP-positive multinucleated cells isolated from RA synovium from resorption pits on dentin slices (37, 38). We therefore used our modified SCID-HuRAg model to conduct a pit formation assay on dentin slices in order to monitor osteoclast function in vivo. The dentin slices were covered with human RA synovium and transplanted into the SCID mice. Pits were not observed at 2 weeks after transplantation, but were observed at 3 weeks after transplantation, indicating that the pits on the dentin slices were probably formed by differentiated osteoclasts.

It is known that human osteoclastogenesis requires human M-CSF and RANKL. Suzuki et al (39) developed an in vitro model of bone destruction using osteoclast-like cells derived from a culture of rheumatoid synovial tissues without any inducers (39). We therefore think that M-CSF and RANKL are provided by the transplanted synovial tissues. Thus, with the use of the new SCID-HuRAg-pit mouse model, we were able to evaluate the suppressive effects of R-125224 on osteoclastogenesis through T cell apoptosis.

Intravenous injection of R-125224 resulted in suppression of osteoclastogenesis in the SCID-HuRAg-pit mouse, as judged from the reduced number of pits on the dentin slices. Histologic analysis of the transplanted tissues revealed that the number of infiltrated lymphocytes around the vessels was decreased after intravenous injection of R-125224. Based on the results of our in vitro and in vivo experiments, R-125224–mediated apoptosis of the infiltrated lymphocytes could be a useful therapeutic strategy for RA in terms of suppressing both inflammation and bone destruction. R-125224 is considered to be a promising therapeutic candidate for RA.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We thank Dr. Masaaki Yoshida (Iwate Medical University) and Dr. Katsumi Sato (Tohoku Rosai Hospital) for providing synovial tissues and Mitsutoshi Uemori (Sankyo Co. Ltd.) for help with the statistical analysis.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES