Tumor Necrosis Factor Alpha Is a Cytotoxin Induced by Murine ...

Vol. 57, No. 5

INFECTION AND IMMUNITY, May 1989, p. 1351-1355 0019-9567/89/051351-05$02.00/0 Copyright X3 1989, American Society for Microbiology

Tumor Necrosis Factor Alpha Is a Cytotoxin Induced by Murine Chlamydia trachomatis Infection DWIGHT M. WILLIAMS,1 2* LYNDA F. BONEWALD,2 G. DAVID ROODMAN,' 2 GERALD I. BYRNE,3 D. MITCHELL MAGEE,'12 AND JULIUS SCHACHTER4 Department of Medicine, Audie L. Miurphy Memorial Veterans Hospital,' and University of Texas Health Science Center,2 San Antonio, Texas 78284; Department of Medical Microbiology, University of Wisconsin Medical School, Madison, Wisconsin 537063; and Department of Laboratory Medicine, University of California at San Francisco, San Francisco, California 941324 Received 30 August 1988/Accepted 19 January 1989

A mouse model of pneumonia caused by murine Chlamydia trachomatis (mouse pneumonitis agent) was used to demonstrate that whole spleen celis from both nude athymic mice (nulnu) and heterozygous mice (nul+) produced tumor necrosis factor alpha in vitro in response to mouse pneumonitis agent antigen. The tumor necrosis factor alpha measured in these supernatants by immunoassay was shown to have bioactivity in a cytotoxic assay in which uninfected target cells were used. This cytotoxicity was distinct from the gamma interferon-related cytotoxicity against C. rachomatis-infected targets that we described previously.

natants was easily detected, without the need to use antiLyt-2 treatment. Furthermore, specific antibody to murine tumor necrosis factor alpha (TFN-a) neutralized the cytotoxicity in the bioassay, and TNF-a was demonstrated to be present in the samples by using a radioimmunoassay (RIA) specific for murine TNF-a.

We have developed a model of murine pneumonia caused by the mouse biovar of Chlamydia trachomatis (mouse pneumonitis agent [MoPn]) (28-30). In this model the athymic nude mouse (nulnu) is significantly more susceptible to primary infection with MoPn than is its heterozygous (nul+) littermate (28-30). In addition, the nul+ mouse can be immunized by sublethal infection with MoPn to become further resistant, while the nulnu mouse cannot (30). We used this model to investigate the production of cytokines by nulnu and nul+ mice in response to MoPn. A cytotoxic factor(s) was produced in spleen cell-conditioned medium (SCM) when spleen cells from C. trachomatis-infected mice were incubated in vitro with C. trachomatis antigen or mitogen (6, 7). Those factors that were produced in response to MoPn antigen were cytotoxic to both C. trachomatisinfected and uninfected fibroblast target cells (6). Antibodycomplement lysis studies showed that the factor(s) responsible for cytotoxicity against MoPn-infected targets was dependent at least in part on T cells (6) for its generation. We found that gamma interferon (IFN--y) was present in SCM from immunized nul+ (but not nulnu or unimmunized nul+) mice (5, 6) and that a neutralizing monoclonal antibody against IFN-y was able to reduce significantly the cytotoxic activity of the SCM. Furthermore, recombinant murine IFN--y was cytotoxic to MoPn-infected targets (6). Thus, IFN--y plays a major role in the cytotoxic activity induced against infected targets. However, SCM also had some cytotoxic activity against uninfected targets which was not reproduced by recombinant murine IFN--y. These data suggest that at least one additional cytokine is present in these samples (6), but the identity of this factor is unclear. In our previous studies, cytotoxicity was only readily demonstrable against uninfected targets after treatment of

MATERIALS AND METHODS Mice. Super-clean nul+ and nulnu mice on a BALB/c background were maintained under barrier conditions (6, 30) and were used for this study. They were free of bacterial and pathogenic viral contamination, as assessed by culture and serology (6, 30). MoPn. The biovar of C. trachomatis was obtained and maintained in embryonated hen eggs as described previously (6, 30). Mice were infected intranasally with 104 inclusionforming units in a volume of 0.05 ml (6). The inoculum was delivered in McCoy modified 5A medium (Difco Laborato-

ries, Detroit, Mich.). MoPn antigen. The antigen used to generate cytokines in SCM was MoPn elementary bodies grown in HeLa 229 cells. It was purified by renografin gradient separation (28); therefore, it contained no egg-derived antigen. Preparation of cells and generation of cytokines. Cytokines were generated as described previously (6). In most studies, unfractionated whole spleen cells (WSCs) from infected or uninfected nul+ or nulnu mice were the cytokine-generating cells. In selected experiments, WSCs were depleted of T cells before cytokine generation by incubating them with anti-Thy-1.2 antibody (Cederlane, London, Ontario, Canada). The WSCs were incubated with anti-Thy-1.2 antibody for 1 h on ice and then were incubated for 1 h at 37°C with a 1:10 dilution of low-toxicity rabbit complement (Pel-Freez Inc., Rogers, Ark.). In two cases (see Table 3), splenic macrophage-enriched preparations were prepared by adhering unfractionated WSCs to sterile petri dishes for 18 h. The plates were washed three times to remove nonadherent cells and were treated with anti-Thy-1.2 and complement to remove T cells. The adherent macrophage-enriched cells were released with a rubber policeman and then used for cytokine generation. WSC suspensions (5 x 106 cells per ml) were incubated with 5 ,ug of UV-inactivated MoPn per ml in

the generating cell population with anti-Lyt-2 and complement. The assay that was used to demonstrate cytotoxicity was a tritiated thymidine release assay with target fibroblasts that were not treated with actinomycin D (6). A chromogenic assay with actinomycin D-treated uninfected fibroblast targets was used in this study. In this assay, cytotoxicity against uninfected targets that was present in the cell super*

Corresponding author. 1351

1352

WILLIAMS ET AL.

a final volume of 20 ml for 48 h in the same medium used previously (5, 6). Macrophage-enriched cells were incubated with MoPn antigen for either 24 or 48 h at 37°C in 5% CO2 in air. Supernatants were collected by centrifugation and were filter sterilized as described previously (6). The supernatants are referred to as WSC-conditioned medium (SCM) and macrophage-enriched conditioned medium, respectively. In selected studies monoclonal anti-mouse IFN-y (hybridoma R4-6A2; Lee Biomolecular, San Diego, Calif.) or anti-TNF-a (graciously supplied by Tim Bringman, Genetech, Inc., South San Francisco, Calif.), by using at least twice the amount of antibody theoretically needed to neutralize the amount of cytokine present (as determined by assay), was added to the SCM- to neutralize IFN-y or TNF-a, respectively. In one experiment (see Table 3), rabbit polyclonal anti-murine TNF-a (Genzyme; Boston, Mass.) was used. In two experiments recombinant murine IFN-y (kindly provided by Genentech) was added to the media in lieu of generated cytokines. Measurement of interferon. Interferon was measured by bioassay with vesicular stomatitis virus and standardized as described previously (5, 6). RIA for murine TNF-a. In the RIA for murine TNF-cx, we used delayed addition of tracer. The RIA was sensitive to 200 pg of murine TNF-ao per ml, was specific for murine TNF-a, and recognized human TNF-a at only microgramper-milliliter concentrations. Murine TNF-a (Genentech) was radiolabeled by the protocol of Bolton and Hunter (4) with '25I-labeled Bolton-Hunter reagent (Amersham Corp., Arlington Heights, Ill.). Murine TNF-a standard was diluted in 0.5 ml of sample buffer (0.15 M NaCl, 0.05 M Tris, 2 mM EDTA, 2 mg of gelatin per ml [pH 7.4]), and 0.5 ml of sample was assayed in duplicate in polystyrene tubes (12 by 75 mm). Anti-murine TNF-a antisera (Genentech) was added to each tube in 100 RI of 2% normal rabbit serum in sample buffer. The samples were incubated overnight at 4°C before 10,000 ppm of 1251-labeled TNF-a in 100 ,ul of buffer was added to each tube. The samples were again incubated overnight at 4°C before the addition of goat anti-rabbit immunoglobulin G (Organon Teknika Corp., West Chester, Pa.) with 15 ,ul of precipitating buffer per ml per tube. The precipitating buffer consisted of phosphate-buffered saline with 5% (wt/vol) polyethylene glycol 6000. The samples were incubated for 1 h at room temperature before they were centrifuged at 2,500 x g for 30 min. The supernatant was aspirated and the pellet was counted in a gamma counter (Quatro; LKB Instruments, Inc., Rockville, Md.). The algorithm used for fitting standards and calculating unknowns was the four-parameter logistic, an iterative weighted linear regression with proper weighting from accumulation of a historical curve (17, 18). TNF-a cytotoxicity assay. The TNF-a cytotoxicity assays were performed essentially as described by Ruff and Gifford (19). L929 murine tumorigenic fibroblasts were seeded at 104 cells per well in 96-well microtiter plates in a 100-,lI volume for 18 to 24 h at 37°C in 5% CO2. The medium was removed, and the cytotoxicity factor was assayed in Eagle minimal

essential medium-5% fetal bovine serum-1 p,g of actinomycin D per ml for 24 h. The medium was removed, and the cells were fixed in 2% Formalin in phosphate-buffered saline for 5 min, washed three times with phosphate-buffered saline, and stained with 0.125% crystal violet in 5% ethanol2% phosphate-buffered saline for 20 min. Samples were run in duplicate. The plate was then washed with distilled water and dried, and 50% ethanol and 1% sodium dodecyl sulfate were added for 30 min at 37°C to dissolve the stain. The plate was read in an enzyme immunoassay plate reader (MA

INFECT. IMMUN. TABLE 1. Neutralization of cytotoxicity by antibody to murine TNF-a Cytotoxicity Sample no.

1 2 3 4 5

Mice and condition

Uninfected nul+ Infected nul+, day Infected nul+, day Infected nul+, day Infected nul+, day

MoPn antigen

10 10 10 10

+ + + +

(U/ml): Without antibody

With antibody

30 ND 140 80 70

NDa ND ND ND ND

"ND, Not detectable.

Bioproducts) at 595 nm. The dilution resulting in 50% survival of cells was determined by interpolation. The number of cytotoxicity units in a sample was defined as the reciprocal of the highest dilution of a preparation that resulted in 50% survival of cells. With TNF-a standards (Genentech), 1 cytotoxicity unit equaled approximately 1012 M TNF-oa. Units were originally calculated per well (100 ,ul) but are reported per milliliter for comparison purposes. Statistics. Results were analyzed by using the Student t test with correction for unequal variances. Results were termed significant when P < 0.05. RESULTS The primary data to be presented in this report were derived by using a RIA specific for murine TNF-'a (see above). Because discrepancies can exist between TNF-a quantitation by antigen determinations by using RIA and biologic activity determination (at least in part because of a faster loss of biologic activity than immunochemical activity [2, 9]), it was also important to demonstrate cytotoxic activity by using a biologic assay. One cytotoxicity unit, as used in this report, is equivalent to approximately 10-12 M murine TNF-a. The correlation between TNF-a molarity and cytotoxicity units (the reciprocal of the highest sample dilution by interpolation resulting in 50% cell survival) was not linear. It was important to demonstrate that the cytotoxicity assay actually measured TNF-a in our spleen cell supernatant samples rather than some other cytotoxic factors. To do this, we used neutralization with specific antibody to murine TNF-a (Genentech). Table 1 shows the results of cytotoxicity assays on five separate samples of SCM from uninfected nul+ mice or nul+ mice infected for 10 days. The first SCM sample was from uninfected WSCs cultured in vitro with MoPn antigen. The second sample was from WSCs from a mouse that was infected for 10 days that was cultured without the addition of MoPn antigen and that had no cytotoxic activity. The last three SCM samples were from infected WSCs that were cultured with MoPn antigen. Cytotoxic activity was present in each of the samples in which antigen was present, and in each sample the activity was completely neutralized by specific antibody to TNF-a. Thus, the biologic activity measured in our cytotoxicity assay was dependent on TNF-a, and the correlation of cytotoxicity with TNF-a used for our assay had relevance for our samples. Table 2 shows the results of measurements by RIA of TNF-a in SCM from uninfected nulnu and nul+ mice and mice infected for 10 days with MoPn. Each result represents an SCM sample from a single mouse run in duplicate. A

TNF-a INDUCED BY MURINE C. TRACHOMATIS INFECTION

VOL. 57, 1989

TABLE 2. Cytotoxicity and TNF-a content of SCM from infected and uninfected mice Mice, condition, and Mce, condtlon, and sample no.

Uninfected nulnu 1 2 3 4

Infected nulnu, day 10 5 6 7 8

TNF-aby (pg/mI;

Cytotoxicity (U/ml; by

antigen

~RIA)

bioassay)

+ +

456 80 1,219 ± 53 439 82 1,273 53

ND"