Main

Many of the adverse events surrounding the birth of a child are associated with inflammation. Maternal infection is associated with cerebral palsy, intraventricular hemorrhage, and white matter damage in both low birth weight and normal birth weight infants (1,2). Chorioamnionitis is associated with preterm birth, neonatal sepsis, neonatal respiratory distress syndrome, intraventricular hemorrhage, cesarean delivery, and a more complicated maternal hospital course (3,4). The genital mycoplasmas, Ureaplasma urealyticum and Mycoplasma hominis, are the most common organisms isolated from amniotic fluid or the chorioamnion in the presence of infection-associated preterm labor (5–7). They are also the most common organisms isolated from the respiratory tracts of preterm infants developing chronic lung disease (8). Many mycoplasmas tend to cause disease, not by directly injuring tissue, but by altering the host's inflammatory response (9). Increased levels of inflammatory mediators in amniotic fluid are also associated with both neonatal and maternal complications (1,10). Stancombe et al. (11) demonstrated that U. urealyticum is capable of stimulating IL-6 and IL-8 production from pulmonary fibroblasts and speculated that this ability might play a role in neonatal chronic lung disease (11). We have shown that U. urealyticum produces an acute inflammatory response in a murine pneumonia model and that hyperoxia potentiates this response (12). We have also reported that this organism is intimately associated with alveolar macrophages in the areas of inflammation (13,14). Two inflammatory mediators produced by macrophages, TNF-α, and NO, are important in regulating inflammation, microbial killing, and tissue injury (15,16). We hypothesized that one mechanism that genital mycoplasmas, U. urealyticum in particular, might cause or potentiate preterm birth and pulmonary inflammation is through the production of inflammatory mediators, such as TNF-α and NO.

Materials. DMEM, L-glutamine, penicillin, and streptomycin were purchased from GIBCO (Grand Island, NY). FCS was obtained from HyClone Laboratories, Inc. (Logan, UT). LPS obtained from Escherichia coli strain 0111:B4, aprotinin, leupeptin, Triton X-100, sodium fluoride (NaF), sodium orthovanidate (Na3VO4), and phenylmethylsulfonyl fluoride were purchased from Sigma Chemical Co. (St. Louis, MO). rIFN-γ was purchased from Genzyme, Corp. (Cambridge, MA), and polymyxin B sulfate was purchased from Pfizer, Inc. (New York, NY).

Organisms. U. urealyticum (serotype 10, isolated from an infant with pneumonia who developed bronchopulmonary dysplasia) and M. hominis (obtained from Gail Cassell, Ph.D., Chair, Department of Microbiology, University of Alabama at Birmingham) were grown to late log phase in 10B medium, aliquoted into 1.5-mL samples, and frozen at -70°C (8). After thawing, the samples contained 3.2 × 107 cfu/mL U. urealyticum and 2.1 × 107 cfu/mL M. hominis. Ureaplasmal serotypes 1, 3, 4, and 8 were obtained from the Perinatal Infectious Disease Laboratory's (Memphis, TN) quality control stocks. Serial 10-fold dilutions were made in DMEM for the inocula. Where appropriate, mycoplasmas were heat-killed by placing them in a 75°C waterbath for 60 min. Quantitative cultures were performed to assure that the organisms had been killed.

Quantitative cultures were obtained by placing 0.1 mL of sample in 0.9 mL of 10B medium and then making serial 10-fold dilutions to 10-5. Twenty microliters of each dilution were then spotted onto A8 agar plates and allowed to dry (17). The 10B broths were incubated at 37°C, and the plates were incubated at 37°C and 5% CO2. The plates and broths were examined daily for 7 d before being discarded as negative.

Macrophages. RAW 264.7 cells were obtained from ATCC (Rockville, MD) and were routinely cultured in DMEM supplemented with 10% FCS, 1 mM L-glutamine, 50 U/mL penicillin G, and 50 µg/mL streptomycin. The cells were split and plated into 24-well microtiter plates and allowed to grow to near confluence. The DMEM medium was removed, 1.0 mL of fresh DMEM was added, and then the cells were used in each experiment. After incubation, the macrophages were >95% viable as determined by the trypan blue exclusion assay.

TNF-α determinations. TNF-α protein levels were determined by ELISA using the Minikit® assay (Endogen, Cambridge MA), which contains a coating antibody, a biotinylated detecting antibody, and murine recombinant TNF-α for a standard. The coating antibody (35 µL) was diluted in 11 mL of PBS, pH 7.4. The detecting antibody (35 µL) was diluted with 11 mL of assay buffer, which consisted of PBS, pH 7.4, 2% BSA (Sigma Chemical Co.), and 0.02% Tween 20 (Sigma Chemical Co.). The horseradish peroxidase-streptavidin (Zymed, South San Francisco, CA) was diluted 1:1200 with the assay buffer. The wash buffer consisted of 50 mM Tris (Bio-Rad Laboratories, Richmond, CA) and 0.2% Tween 20 at pH 7.4. The stop solution was 0.18 M H2SO4. In brief, this assay uses a sandwich ELISA technique specific for murine TNF-α. The instructions provided with the kit were followed without deviation. The values were reported in picograms/mL of culture supernatant. Each sample was run in duplicate and read on an HTII microplate reader at 450 nm (Anthos Labtec, Inc., Frederick, MD), and the average was plotted against a standard curve using Deltasoft® software (Biometalics, Inc., Princeton, NJ) for Macintosh computers. The limit of detection for this assay was <50 pg/mL.

Immunoblotting. Accumulation of iNOS protein was determined by immunoblotting. Cells were lysed in extraction buffer (20 mM Tris, 100 mM NaCl, 1% Triton X-100, 50 mM NaF, 1 mM Na3VO4, 0.2 mM phenylmethylsulfonyl fluoride, 10 µg/mL leupeptin, and 10 µg/mL aprotinin). Lysates were electrophoresed on 7.5% SDS-polyacrylamide gels and transferred to nitrocellulose membranes. Blots were reacted with a murine MAb specific for iNOS expressed in macrophages, p130macnos (Transduction Labs, Inc., Lexington, KY), followed by a sheep anti-mouse IgG horseradish peroxidase conjugate (Amersham, Arlington Heights, IL). Proteins were detected by enhanced chemiluminescence (Amersham ECL).

Experimental design. Blank wells of a 24-well microtiter plate, containing approximately 106 macrophages and 1.0 mL of DMEM, were inoculated with 0.1 mL of sterile 10B medium, whereas sample wells were inoculated with 0.1 mL of 10B broth containing mycoplasmas to obtain 104, 105, and 106, cfu/mL per well, n = 10. In four experiments 107 cfu/mL per well was also used to study very high organism concentrations. LPS was added to the appropriate wells to final concentrations of 1, 10, and 100 ng/mL. Experiments examining the effects of different serotypes were performed 12 times. In experiments designed to block LPS-like activity, polymyxin B sulfate was added to the macrophages 1 h before the experiments began at a final concentration of 10 µg/mL, n = 8. In experiments examining the effects of rIFN-γ, a final concentration of 10 U/mL was used, n = 8. After inoculation, the plates were placed in a 37°C CO2 incubator for 16 h. The supernatant was then removed and frozen at -70°C until assayed. Experiments studying either TNF-α production or organism viability over a 24-h period were performed in 6-well microtiter plates, with a final volume of 5.0 mL. Sampling, 0.2 mL, was performed with replacement of fresh DMEM at 0, 2, 4, 8 and 24 h, n = 2.

For iNOS determinations, 6-well microtiter plates were inoculated with 5.0 × 106 macrophages in 5.0 mL of DMEM per well, n = 3. The cells were allowed to adhere for approximately 4 h before each experiment. The DMEM and nonadherent cells were removed, and 5.0 mL of fresh DMEM were added to each well. Wells were inoculated with either LPS (1.0 or 100 ng/mL), U. urealyticum (104, 105, or 106 cfu/mL), or M. hominis (104, 105, or 106 cfu/mL) either by themselves or with rIFN-γ (1.0 or 10 U/mL). The cells were allowed to incubate at 37°C and 5.0% CO2 for 16 h. The cells were then harvested for assay as described above.

Statistics. The Kruskal-Wallis ANOVA by ranks median test (Statistica Mac, Statsoft, Inc., Tulsa, OK) was used to test for statistical significance. The level of significance was set at p < 0.05.

RESULTS

The ability of U. urealyticum and M. hominis to stimulate TNF-α production from murine macrophages is demonstrated in Figures 1 and 2. TNF-α production increased in a stepwise fashion with increasing amounts of LPS, whereas only small amounts of TNF-α were produced in unstimulated macrophages. The effect of sterile 10B broth medium was determined by making a 1:10 and 1:100 dilution of sterile 10B broth medium in DMEM and inoculating the plates. The 1:10 dilution produced marginal amounts of TNF-α, mean of 735 pg/mL, whereas the 1:100 dilution produced a mean of only 341 pg/mL, neither of which was different from baseline (p > 0.05). TNF-α production increased in a stepwise fashion from 104 through 106 cfu/mL of U. urealyticum, whereas an increase in TNF-α production was noted from 104 to 105 cfu/mL of M. hominis, but at higher concentrations a significant decline was noted (Fig. 1). All serotypes of U. urealyticum tested produced significant amounts of TNF-α (Fig. 2).

Figure 1
figure 1

TNF-α production by genital mycoplasmas. Both LPS and U. urealyticum significantly stimulate TNF-α production in a dose-dependent fashion. At the higher concentrations of M. hominis, TNF-α production was inhibited (p < 0.01). Each data point represents the average, ± SEM, of 10 experiments with the exception of the 107 concentrations, which were performed four times each. *p < 0.05, **p < 0.01, ***p < 0.001.

Full size image
Figure 2
figure 2

TNF-α production by ureaplasmal serotypes. All serotypes of U. urealyticum tested, representative of both human ureaplasmal serovars, produced significant amounts of TNF-α (p < 0.001). Variation in TNF-α production was seen between the different isolates, but this did not reach statistical significance (p > 0.05). Each data point represents the average, ± SEM, of 12 experiments.

Full size image

Pretreatment of RAW 264.7 cells with polymyxin B sulfate before experimentation significantly reduced, but did not eradicate, the TNF-α production associated with LPS but had little effect on TNF-α production associated with either U. urealyticum or M. hominis (Fig. 3). Coincubation of RAW 264.7 cells with rIFN-γ slightly increased TNF-α production in the blank wells, but little effect was seen with either mycoplasmal isolate on TNF-α production (Fig. 4). A decline in TNF-α production was seen with the macrophages coincubated with rIFN-γ and stimulated with LPS (Fig. 4).

Figure 3
figure 3

Effect of LPS blockade on TNF-α production. Polymyxin B (10 µg/mL) significantly reduced TNF-α production in response to 10 and 100 ng/mL of LPS (*p < 0.05 and **p < 0.01, respectively) but had no effect on the TNF-α response to the genital mycoplasmas (p > 0.05). The data points represent the average, ± SEM, of eight experiments.

Full size image
Figure 4
figure 4

Effect of IFN-γ on TNF-α production. Interferon-γ (10 U/mL) tended toward an increase in TNF-α production in the blank and mycoplasmas, but this reached statistical significance only in the blank (p < 0.01). Interferon-γ was associated with reduced TNF-α production in the LPS samples (p < 0.05). The data points represent the average, ± SEM, of eight experiments.

Full size image

To test whether the mycoplasmal component responsible for TNF-α production was heat-labile or not, experiments were performed with heat-killed organisms (Fig. 5). No difference in TNF-α production was found between the blank and heat-treated 10B medium (p > 0.05). Heat-killed U. urealyticum were significantly hampered in their ability to elicit a TNF-α response at the 104 and 105 concentrations (p < 0.001) but not at the 106 concentration (p > 0.05) (Fig. 5). The results with heat-killed M. hominis were dependent on the concentration and the effect was not as large compared with U. urealyticum. Heat-killed organisms were unable to produce as much TNF-α compared with viable organisms at the 104 concentration (p < 0.05), but at the higher concentrations they were able to produce as much or more TNF-α as were live organisms (Fig. 5).

Figure 5
figure 5

TNF-α production by heat-killed organisms. Heat-killed U. urealyticum had a significantly reduced capacity to stimulate TNF-α production compared with viable organisms (p < 0.001) at the 104 and 105 concentrations. A small reduction in the capacity of heat-killed M. hominis to stimulate TNF-α production was found at the 104 concentration (p < 0.05), none at 105, and an increase at 106 (p < 0.01). The data points represent the average, ± SEM, of 13 experiments.

Full size image

TNF-α production was detectable by 2 h of incubation with LPS, U. urealyticum, and M. hominis and continued to increase to 24 h (Fig. 6). Again, the 106 dilution of M. hominis tended to produce less TNF-α than the 105 dilution starting at about 4 h of incubation. Quantitative culture of the cell supernatants at each time point revealed a steady decline in the number of viable organisms such that no viable organisms remained at 24 h of incubation (Fig. 7). Organism death was expected because DMEM is not adequate for genital mycoplasmal growth. Macrophages remained greater than 90% viable throughout these experiments as determined by the trypan blue exclusion method.

Figure 6
figure 6

Kinetics of TNF-α production over 24 h. This graph depicts TNF-α production during a 24-h period. Samples of 0.2 mL were obtained with replacement, at 0, 2, 4, 8, and 24 h. Very little TNF-α production was noted at the 0- and 2-h time points; thereafter, increases were noted that correlated with the concentration of organisms used. Differences in TNF-α production between the 104 and 105 cfu/mL concentrations were discernable by 4 h of incubation. Data points respresent the mean of two experiments.

Full size image
Figure 7
figure 7

Viability of organisms over 24 h. This graph depicts the quantity of viable organisms remaining at 0, 2, 4, 8, and 24 h after inoculation of the macrophages. Viability seemed to decline precipitously when only 105 cfu/mL organisms remained. All organisms were dead by 24 h of incubation. These time points correspond to the TNF-α values shown in Figure 6. Notice that TNF-α production continues even though the organisms are dying. Data points represent the mean of two experiments.

Full size image

Neither M. hominis (Fig. 8A) nor U. urealyticum (Fig. 8B) alone triggered detectable iNOS protein accumulation in RAW 264.7 macrophages, but both organisms up-regulated iNOS protein production in a dose-dependent manner in these cells in the presence of low concentrations of rIFN-γ (Fig. 8). However, exposure of RAW 264.7 cells to the highest tested concentration (106 cfu/mL) of M. hominis but not U. urealyticum was consistently associated with reduced iNOS protein accumulation.

Figure 8
figure 8

iNOS production by mycoplasmas. Both M. hominis and U. urealyticum trigger iNOS protein accumulation by RAW 264.7 macrophages. Cells were cultured for 18 h in medium alone, with the indicated concentrations of LPS (control), or with increasing concentrations of M. hominis (A) or U. urealyticum (B) with or without rIFN-γ at the indicated concentrations. After incubation, cells were lysed and iNOS protein levels were determined by immunoblotting as described in "Methods."

Full size image

DISCUSSION

These data demonstrate that genital mycoplasmas stimulate the production of iNOS, in the presence of rIFN-γ, and TNF-α from murine macrophages in a dose-dependent fashion, and that viable organisms are not needed for this effect. At the highest concentrations of M. hominis, a decline in both TNF-α and iNOS production was noted. One possible mechanism for this reduction in TNF-α production could be macrophage injury, possibly by the generation of toxic oxygen species by M. hominis, although significant membrane injury could not be found with trypan blue exclusion (18). Occasionally, this effect was also seen with U. urealyticum that does not produce toxic oxygen species. Although the reduction in TNF-α and iNOS production found at higher concentrations of M. hominis is interesting, we cannot draw any conclusions at this time because, in preliminary experiments, another M. hominis strain (ATCC 23114, serotype V) did not produce this effect (data not shown). Cytokine production from mycoplasmal-stimulated inflammatory cells has been demonstrated by other investigators (19–22). Stancombe et al. (11) demonstrated that U. urealyticum induces the proinflammatory cytokines, IL-6 and IL-8, from neonatal fibroblasts and postulated that ureaplasmal-stimulated cytokine release from fibroblasts might contribute to the development of bronchopulmonary dysplasia. Because the macrophage and fibroblast play such an important role in inflammation and repair, the induction of inflammatory mediators by these microorganisms is likely a significant factor in the pathogenesis of mycoplasmal disease.

Although both LPS and genital mycoplasmas can elicit the production of TNF-α from murine macrophages, the mechanism does not appear to be the same. Polymyxin B blocks TNF-α production by LPS but has no effect on TNF-α production in response to the genital mycoplasmas. Similar results have been obtained with partially purified membrane lipoproteins from Mycoplasma arginini (23,24). Also, Mycoplasma capricolum membranes and intact Mycoplasma mycoides ssp mycoides stimulate TNF-α production from both the LPS-responsive C3H/HeN and LPS-unresponsive C3H/HeJ-derived macrophages, whereas TNF-α is produced by LPS only in the C3H/HeN-derived macrophages (25). Therefore, our data confirm that, like other mycoplasmas, the genital mycoplasmas stimulate TNF-α production by a pathway that differs from that of LPS.

The importance of the mycoplasmal membrane molecular structure is now only starting to be understood. Sher et al. (26) have demonstrated with M. capricolum that the ability to stimulate TNF-α from macrophages is associated exclusively with the plasma membrane. Rosendal et al. (25) demonstrated the moiety in M. mycoides ssp. mycoides that induces TNF-α production is heat-tolerant. Our findings with whole organisms demonstrate that the M. hominis moiety that elicits TNF-α production in murine macrophages is more heat-tolerant than the ureaplasmal moiety.

Other investigators have shown that the release of TNF-α, IL-1β, and toxic oxygen species, and production of iNOS is enhanced by activating macrophages with rIFN-γ (25,27–29). Our data show that rIFN-γ only slightly enhances the production of TNF-α in response to challenge of RAW 264.7 murine macrophages by genital mycoplasmas similar to what Rosendal et al. (25) found with Mycoplasma mycoides spp mycoides. Neither M. hominis nor U. urealyticum alone stimulated iNOS protein production. However, both organisms triggered iNOS protein accumulation in a dose-dependent fashion in the presence of low concentrations of rIFN-γ, paralleling our previously published results with Gram-positive bacterial lipoteichoic acid (30).

The expression of iNOS protein by human macrophages and its potential role in human disease has been controversial (31–35). However, recent studies have clearly shown that human peripheral blood monocytes and alveolar macrophages are capable of producing iNOS mRNA, protein, and enzyme activity (36,37). Anstey et al. (37) detected iNOS protein in peripheral blood mononuclear cells isolated from Tanzanian children and found that the high output NO pathway was associated with protection from severe malaria. Nicholson et al. (36) found that 65% of the alveolar macrophages isolated from tuberculosis patients expressed immunoreactive iNOS protein and enzyme activity (compared with 10% of alveolar macrophages from control patients). These data provide incontrovertible evidence for iNOS expression in human myeloid cells and suggest that the high output NO generation pathway catalyzed by iNOS may play an important role in host defense.

The unique biologic properties of mycoplasmas make them ideal pathogens in the perinatal period (9). Unfortunately, because of their ubiquitous nature, difficulty in culturing, and frequent coisolation with other bacteria, mycoplasmas are often overlooked as human pathogens (38). Similar problems are found in animal infections with mycoplasmas; but because of the ability to control breeding colonies and inoculate gnotobiotic animals, mycoplasmas are now known to cause chronic pneumonias in calves, lambs, pigs, chicks, turkeys, quail, rats, and mice (39,40). Significantly, many of these diseases are subclinical unless a co-factor, such as another infectious organism or air pollutant is present (12,39). Mycoplasmal genital tract disease, including salpingitis, endometritis, reduced breeding efficiency, vulvovaginitis, and vertical transmission to offspring is also prevalent in animals (39,40). In humans the disease processes are analogous. Genital mycoplasmas cause neonatal pneumonia, persistent pulmonary hypertension, abscesses, septic arthritis, endometritis, pelvic inflammatory disease, postpartum fever, and chorioamnionitis (14,38,41). The genital mycoplasmas are also associated strongly with spontaneous abortion, preterm birth, and neonatal chronic lung disease (41,42). As in many animals, mycoplasmas colonize the human female lower genital tract where they cause no apparent harm (43). It is only when they gain access to the upper genital tract, fetus, or neonate that they are associated with disease (43).

Inflammation is an important pathologic process in the perinatal period. Elevations in maternal proinflammatory mediators are associated with preterm birth, chorioamnionitis, maternal fever, and endometritis; whereas elevations in neonatal proinflammatory mediators are associated with neonatal sepsis, pneumonia, intraventricular hemorrhage, and chronic lung disease (44). Our data demonstrate that the genital mycoplasmas, which are one of the most common organisms isolated from either the mother or infant, are capable of stimulating the production of proinflammatory mediators from murine macrophages. These organisms should not be discounted as potential pathogens in the perinatal period.