Research Paper Mediators of Inflammation, 10, 199–208 (2001)

TUMOR necrosis factor-alpha (TNF-alpha) plays an important role in the pathogenesis of rheumatoid arthritis. The present study was to evaluate the effects of lipopolysaccharide (LPS), phytomitogens and cytodifferentiation agents on cytotoxicity of TNF-alpha secreted by adherent human mononuclear cells (AMC). TNF-alpha cytotoxicity in LPS-treated, phytomitogen-treated, and cytodifferentiation agent-treated AMC supernatants were analyzed by the L929 bioassay system. Our results showed that LPS could induce homogeneous TNF-alpha production by AMC whereas, in addition to TNF-alpha, phytomitogens could also induce other TNF-like factors. Neither methotrexate, retinoic acid nor sodium butyrate can inhibit TNF-alpha cytotoxicity, while hexamethylene bisacetamide could not only inhibit TNF-alpha cytotoxicity but also TNF-alpha inducing ability of LPS to AMC.


Introduction
Rheumatoid arthritis (RA) is a chronic autoimmune disease with clinical features of inflammation and destruction of joints. The pathogenic mechanism of RA is still unknown. Several inflammatory mediators and cytokines have been identified in the synovial fluid of RA patients. 1 These mediators play important roles in RA pathogenesis since they can induce the release of tissue-damaging enzymes through cell-cell interaction and result in inflammation and destruction of joints.
Tumor necrosis factor-a (TNF-a ) is a pleiotrophic cytokine secreted mainly by monocytes and macrophages. TNF-a has many overlapping bioactivities with interleukin 1 (IL-1). Some of these bioactivities, such as induction of synovial fibroblast proliferation, 2 stimulation of fibroblasts and chondrocytes to secrete prostaglandin E 2 (PGE 2 ) and collagenase, 3 stimulation of monocytes, endothelial cells and cultured rheumatoid synovial cells to produce IL-1, 4 and activation of polymorphic multinuclear leukocytes, 5 were reported to be associated with the inflammation, degeneration and destruction of joints. TNF-a in synovial fluid of RA patients was synthesized by synovial tissue and correlated with their seropositivity of TNF-a and disease activity. 6 High levels of TNF-a mRNA were detected on rheumatoid synovial membrane. 7 According to these observations, TNF-a indeed plays a crucial role in RA pathogenesis.
Non-steroid anti-inflammatory drugs, penicillamine, azathioprine and cyclophosphamide are usually prescribed for RA treatment; however, these effective medications can lead to adverse side effects. Several studies have demonstrated that a low dose of methotrexate (MTX) was an effective medication in RA intervention with very limited toxicity. 8 The antiinflammatory capacity of MTX was postulated to be contributed to effective RA treatment. MTX can inhibit not only the synthesis of immunoglobulin M-rheumatoid factor, but the regeneration of methionine to block the synthesis of two important mediators in cellular immune responses: S-adenosylmethionine (SAM) and polyamine. 9 However, the exact mechanism of MTX to inhibit RA progress remained unclear. Segal et al. has reported that MTX can inhibit neither the synthesis nor secretion of IL-1 both in in vivo and in vitro systems. 10 Our present study focused on the elucidation of MTX effects on the cytotoxicity of TNF-a secreted by adherent mononuclear cells (AMC). In addition, our study was the first to systemically analyze and demonstrate the effect of various cytodifferentiation agents that drive HL-60 cells differentiation, such as retinoic acid (Ra), sodium butyrate (SB), and hexamethylene bisacetamide (HMBA), on the cytotoxicity of TNF-a secreted by AMC.
The selection of a simple and rapid bioassay with characteristics of high sensitivity and good reproducibility is important for the analysis of TNF-a cytotoxicity. According to the literature, the standard error of the sample mean (SEM) in the standard L929 bioassay system for TNF-a quantitation was reported to range from 5 to 15%. The average SEM of the L929 cytotoxicity bioassay system established in the present study could be reduced to be less than 5% (on average, 2-3%) with our modification. Therefore, our modified L929 bioassay system was rather consistent and would be of benefit to the analysis or quantitation of TNF-a cytotoxicity.

AMC isolation
Peripheral blood mononuclear cells (PBMC) were isolated by Ficoll-Hypaque gradient centrifugation as described previously. Briefly, 30 ml of fresh peripheral blood was gently added onto the top of 24 ml of Ficoll-Hypaque solution. Buffy coat was cautiously collected after centrifugation at 450 ´g for 30 min at 20°C and washed three times with Hank's balanced salt solution (HBSS). Ten milliliters of PBMC (4 ´10 6 cells/ml) was incubated at 37°C for 1.5-2 h in a 100 mm cell culture plate and washed vigorously (shaking at 150 r.p.m. for 2 min) three times with HBSS. Ice-cold complete RPMI-1640 medium (RPMI-1640 with 10% fetal bovine serum, 2 mM of L-glutamine, 100 U/ml of penicillin and 10 m g/ml of streptomycin) was added to the cell, incubated for 30 min on ice, and AMC were collected by rubber policeman. Cell density was adjusted to 5 ´10 5 cells/ ml using complete RPMI-1640 medium. Cell survival rate and the percentage of AMC were determined as > 98% and > 94% by trypan blue exclusion test and nonspecific esterase stain, respectively.

AMC treatment
One milliliter of AMC (5 ´10 5 cells/ml) was cultured in each well of a 24-well tissue culture plate and incubated at 37°C for 1 h. The medium was cautiously discarded after the AMC had adhered to the culture plate. TNF-a secretion of AMC cells was analyzed by adding 1 ml of RPMI-1640 containing various concentrations of phytomitogens and cytodifferentiation agents, respectively, in the presence or absence of lipopolysaccharide (LPS) activation at 37°C (summarized in Table 1). Culture medium was collected after desired intervals, centrifuged at 500 ´g for 5 min at   stored at -80°C until use. The survival rate of adherent cell was determined to be > 95% after the culture medium was collected. In investigating the dose effect of MTX and cytodifferentiation agents on the TNF-a secretion by AMC, 5 ´10 5 cells/well of AMC cells were incubated with 1 ml of LPS (0.02 m g/ml) containing RPMI-1640 with MTX (1 ng/ml-1 mg/ml), Ra (0.5-80 m M) or SB (5-800 mM), respectively, at 37°C. For ruling out the possibility that there might be some mutual interaction between MTX or cytodifferentiation agents and LPS, LPS-containing medium was removed after 1 h of pulse and AMC was washed three times with serum-free medium before treatment with MTX or cytodifferentiation agents. Culture supernatants were collected and stored as already described.

Cell culture
Murine L929 fibroblasts and human HEp-2 cells were incubated in RPMI-1640 complete medium (RPMI-1640 with 10% fetal bovine serum, 2 mM of L-glutamine, 100 U/ml of penicillin and 100 m g/ml of streptomycin) and DMEM medium, respectively, at 37°C.

Evaluation of cellular susceptibility to recombinant human TNF-a
Confluent monolayer murine L929 fibroblasts were trypsinized and resuspended to 2 ´10 4 cells/ml with RPMI-1640. One hundred microliters of cell suspension was seeded to each well of a 96-well tissue culture plate. The medium was discarded after overnight culture and replaced by RPMI-1640 containing various concentrations of recombinant human TNF-a (rhTNF-a ) and 1 m g/ml of actinomycin D. MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide, tetrazolium salt) assay was conducted after 16-20 h of culture at 37°C. The sensitivity of HEp-2 to rhTNF-a was analyzed using similar procedure without actinomycin D exposure.

L929 cytotoxicity bioassay
L929 cytotoxicity was analyzed by modifying the methods form Baarsch et al. 11 and Galloway et al. 12 The flowchart of our modified version is summarized in Fig. 1 The mean ± SEM of the percentage of L929 cytotoxicity was calculated by data obtained from three to eight wells.

Neutralization test
The neutralization test was conducted according to Galloway et al. 12 Briefly, AMC supernatants were diluted by RPMI-1640 and co-incubated with equal volume of various concentration of TNF-a monoclonal antibody solution containing 2 m g/ml actinomycin D at room temperature for 2 h. One hundred microliters of the diluted supernatant was added to each well of a 96-well tissue culture plate and the L929 cytotoxicity test was conducted.

Statistical analysis
The statistical significance of the mean ± SEM was analyzed by Student's t-test.

Results
The present study focused on using a modified L929 cytotoxicity bioassay system for TNF-a quantitation to analyze the effect of MTX and cytodifferentiation agents, such as Ra, SB, and HMBA, on the cytotoxicity of TNF-a secreted by AMC. In addition, the effect of LPS, 4b -phorbol-12b -myristate-13a -acetate (PMA), phytohemagglutinin (PHA), pokeweed mitogen (PWM) and concanavalin A (ConA) on cytotoxicity of TNF-a secreted by AMC were also investigated.
In traditional MTT cytotoxicity assay, the tetrazolium salt will be reduced to produce Formazan crystals by mitochondrial succinate dehydrogenase in viable cells. TNF-a cytotoxicity could be indirectly examined by enzyme-linked immunosorbent assay reader after the Formazan crystals were completely dissolved by solvents. Therefore, whether the crystals could be completely dissolved would significantly affect the result of cytotoxicity test. Among various solvents reported in the literature, 10% sodium dodecyl sulfate (SDS)-0.01 N HCl, 10% SDS-50% isopropanol, 20% SDS-50% dimethylformamide, 10% SDS-50% DMSO, and DMSO alone were relatively better solvents for Formazan crystal-dissolving purpose (Table 2). In terms of the incubation time and mixing requirement, DMSO was the most convenient solvent. Therefore, we chose DMSO as the crystaldissolving solution in our study.
HEp-2 and L929 are cell lines that have been commonly used for the analysis of TNF-a cytotoxicity. We first tested the susceptibility of those two cell lines to rhTNF-a for establishing the bioassay system. The 50% inhibition concentration (IC 50 ) of TNF-a to the HEp-2 cell was determined to be higher than 10 ng/ml ( Fig. 2A), while that of the L929 cell was approximately 0.18 ng/ml (Fig. 2B). Therefore, we chose L929 cells as target cells in our assay system since L929 cells exhibited relatively high sensitivity to TNF-a and might perform better precision in the evaluation of TNF-a cytotoxicity.
AMC was cultured in complete RPMI-1640 medium containing various agents respectively listed in Table  1. TNF-a cytotoxic activity of the AMC cell-free culture supernatants under the exposure of various agents was analyzed by L929 bioassay system. The results are summarized in Table 1. MTX and cytodifferentiation agents showed no stimulatory capacity on AMC to produce TNF-a or TNF-like factors (< 10% of TNF-like cytotoxicity to the L929 cell). Among the agents tested in the study, ConA and PMA were relatively the strongest and weakest that could induce the TNF-like cytotoxicity, respectively.
Under continuous LPS exposure, AMC could be immediately activated and the TNF-a cytotoxicity was rapidly increased and reached a plateau within 10 h. Therefore, high TNF-a cytotoxicity could be detected in AMC culture supernatants by L929 cytotoxicity assay (Fig. 3A). When AMC was temporarily pulsed by LPS for only 1 h, residual TNF-a cytotoxicity could be consistently detected after 36 h of LPS removal; however, the cytotoxicity would be significantly Mediators of Inflammation · Vol 10 · 2001 decreased thereafter (Fig. 3B). For identifying whether LPS could induce AMC to secrete other factors that exhibited TNF-like cytotoxicity, various concentrations of anti-TNF-a monoclonal antibody (5-100 ng/ml) was added into the cell-free supernatants to neutralize TNF-a cytotoxicity, and the residual TNF-like cytotoxicity of the AMC supernatant was subsequently analyzed. Fig. 4 shows that the TNF-like cytotoxicity would be completely inhibited when the concentration of anti-TNF-a monoclonal antibody was higher than 20 ng/ml. The results demonstrated that TNF-like cytotoxicity induced by LPS was mainly contributed by homogeneous TNF-a ( Fig. 4 and Table 3). However, in addition to TNF-a , the TNF-like cytotoxicity induced by other mitogens, such as PHA, PMA, and ConA, seemed to be caused by factors other than TNF-a since there was residual TNF-like cytotoxic activity after anti-TNF-a monoclonal antibody neutralization (Table 3). We next analyzed the effect of MTX and cytodifferentiation agents on cytotoxicity of TNF-a secreted by LPS-activated AMC. AMC were activated with 0.02 m g/ml LPS in the presence of MTX at various concentrations (1 ng/ml-1 mg/ml), TNF-a cytotoxicity of the culture supernatants was analyzed after overnight incubation. Our results showed that MTX, either in the condition of continuous (p > 0.1; Fig. 5A) or temporary (p > 0.1; Fig. 5B) exposure, had no inhibitory capacity to cytotoxicity of TNF-a secreted by AMC. The cytotoxicity of rhTNF-a activity would neither be affected when rhTNF-a was incubated overnight with various concentration of MTX (data not shown). For ruling out the possibility that lack of MTX inhibition to cytotoxicity of TNF-a secreted by AMC was due to mutual physical or chemical interaction between LPS and MTX, TNF-a cytotoxicity in the supernatants was analyzed after AMC was first pulsed with LPS for 1 h and subsequently treated with MTX for 2 or 16 h after LPS removal. The results showed that MTX remained no inhibition capacity to TNF-a cytotoxicity (Fig. 6A,B).
The residual mitogens and cytodifferentiation agents in the AMC culture supernatants might cause cytotoxic effect that would reinforce the TNF-a cytotoxicity. Therefore, we further investigated the L929 cytotoxicity caused by each agent for ruling out this possibility. These agents were proved not to exhibit cytotoxicity to L929 cells under the concentration of AMC treatment (< 0.5 U/ml; data not shown). The cytotoxicity of rhTNF-a was also not inhibited by those agents when we incubated L929 cells in the presence of each agent and rhTNF-a . In addition, 1 ng/ml to 2 m g/ml of anti-TNF-a monoclonal antibody had no L929 cytotoxicity either (data not shown). Only high dosage of HMBA (> 40 mM) showed substantially low cytotoxicity (1-1.5 U/ml) among cytodifferentiation agents (data not shown). Fig. 7 demonstrates the results of the effect of Ra, SB, and HMBA to cytotoxicity of TNF-a secreted by AMC. The data showed that only HMBA (0.5-80 mM) had significant inhibition capacity to cytotoxicity of TNF-a secreted by LPS-activated AMC, whereas Ra (0.5-80 m M) and SB (5-400 m g/ml) showed no inhibition (p > 0.05).
For ruling out the possibility of mutual interaction between LPS and cytodifferentiation agents and further investigating the correlation between cytodifferentiation agents and TNF-a cytotoxicity, TNF-a activity was analyzed after AMC was first pulsed by LPS for 1 h and subsequently incubated with cytodifferentiation agents. Fig. 8 shows that only high concentrations of HMBA (> 10 mM) and SB (> 400 m g/ml) exhibited significant and slightly inhibitory capacity to cytotoxicity of TNF-a secreted by LPS-activated AMC. Similar results were obtained when the HMBA was removed after overnight incubation with LPSactivated AMC to allow further TNF-a secretion (data not shown).

Discussion
Many cell lines have been used in the analysis of TNF-a cytotoxicity. The IC 50 of rhTNF-a (with specific activity (2-4) ´10 8 U/ml) to ME-180, HeLa and MRC-5 cells was approximately 1 ng/ml, and that of U937 cell was about 0.3 mg/ml. 12 A simple TNF-a bioassay was established using HEp-2 cells as target cells; 13 however, the IC 50 of rhTNF-a to HEp-2 cells was detected to be higher than 10 ng/ml in this system ( Fig. 2A). This result supported the observation that HEp-2 cells showed low sensitivity to TNFa . WEHI 1640 subclone 13 cells were reported to be more sensitive to TNF-a than L929 cells, but the cell density was too high and their sensitivity to TNF-a would be decreased or even turned out to be insensitive after multiple subcloning. 14 The L929 cell line was the most common target cell for the detection of TNF-a cytotoxicity 15 since the analysis of TNF-a cytotoxicity to L929 would not be affected by rhIL-1 and interferon-a . Therefore, we chose L929 as target cells in the analysis of TNF-a cytotoxicity. The chemicals used in our study had no cytotoxic effects to L929 and the IC 50 of rhTNF-a to L929 cells was only about 0.18 ng/ml in our study (Fig. 2B).
Conventional analytic methods for cellular growth/ proliferation and survival included calculation of   16 Cell survival can be evaluated by measuring the absorbance of Formazan crystals dissolved in solution at 570 nm. The number of survival cells and the cytotoxicity of TNF-a is proportional and reciprocally correlated with the absorbance. The SEM of MTT colorimetric assay of TNF-a cytotoxicity on L929 cells was reported to be among 5-15%. 15,17,18 But the SEM of MTT test has been reduced to be less than 5% (on average, 2-3%) by our modification. Our modifications of MTT assay were as following (Fig. 1). L929 cells were allowed to be pre-incubated overnight for better stability. The initiative cell density was 2 ´10 4 cells/well in a 96-well tissue culture plate, and this would become approximately 95% confluence when harvested. Control of cell number was an important factor that could remain constant TNF-a sensitivity of the target cells. Besides, the removal of culture medium before Formazan crystal dissolution by centrifugation could be of benefit to both the prevention of a pseudopositive result due to phenol red in medium and a significant increase in the dissolution of the Formazan crystal. The results from our modified MTT assay had low SEM and high reproducibility. We expected that this modified L929 bioassay system would be of great potential to be applied in clinical analysis or quantitation of TNF-a cytotoxicity. The effect of PMA, PHA and ConA to PBMC has been investigated, 19 and the results indicated that mitogen was a relative stronger inducer of TNF-a secretion on PBMC. Our study investigated the induction ability of LPS and other mitogens, such as PHA, PMA, ConA and PWM, on cytotoxicity of TNF-a secreted by AMC. We found that LPS could induce homologous TNF-a secretion; in addition, its induction ability was higher than that of other mitogens ( Table  2). The nature and characteristics of TNF-like factors induced by mitogen-induced AMC await further study to be elucidated.
MTX is an effective chemical in the RA intervention. Nevertheless, its mechanism of action needed further investigation. MTX exhibited anti-inflammatory activity 8 -10 such as inhibition of PMN and monocyte chemotaxis, 20 inhibition of B-cell and T-cell function, 21 and indirect inhibition of SAM and polyamine formation through downregulation of dihydrofolate reductase. Segal et al. 10 demonstrated that MTX could selectively inhibit IL-1 activity rather than its synthesis and secretion. The present study examined the effect  Table 1). After overnight culture, cell-free supernatants were harvested and TNF cytotoxicity in 10 ´diluted supernatants was analyzed.  Table 1). TNF cytotoxicity in 5 ´diluted supernatants was analyzed.
of MTX or cytodifferentiation agents on the cytotoxicity of TNF-a secreted by LPS-activated AMC. We found MTX (1 ng/ml-1 mg/ml) showed no inhibition to TNF-a cytotoxicity (Figs. 5 and 6) whereas HMBA exhibited significant inhibitory effect on cytotoxicity of TNF-a secreted by LPS-activated AMC (Fig. 7). The inhibition of HMBA to cytotoxicity of TNF-a secreted by LPS-activated AMC might be due to the following mechanisms. First of all, HMBA could kill AMC since higher concentrations of HMBA exhibited cytotoxicity to L929 cells. Second, the mutual interaction between HMBA and LPS could block AMC activation by LPS. Third, HMBA could indeed inhibit TNF-a synthesis and secretion. Forth, HMBA can inhibit TNFa cytotoxicity to L929. The first possibility can be ruled out by the trypan blue exclusion test. High concentrations of HMBA (> 10 mM) indeed exhibited an inhibitory effect on TNF-a cytotoxicity to L929 (Figs. 7 and 8). HMBA also could partially block AMC activation by LPS (Figs. 7 and 8). However, whether the TNF-a sensitivity of L929 could be decreased by HMBA awaits further study.
On analysis of various cytodifferentiation agents and mitogens, we found HMBA showed inhibitory effect of TNF-a cytotoxicity of L929. However, whether HMBA could inhibit other biological effects of TNF-a , such as the stimulation of fibroblast proliferation, induction of PGE 2 and collagenase, and be a potential medication for RA treatment requires further investigation. Besides, the activation of CD4 + T cells is suggested as an important factor in RA pathogenesis; 22 therefore, it would be intriguing to examine whether MTX and HMBA could regulate other RA pathogenesis-related cytokines.