Synergy between tumor necrosis factor and bacterial products causes hemorrhagic necrosis and lethal shock in normal mice ( Shwartzman phenomenon / endotoxin shock / lipopolysaccharide / mycoplasma / corynebacterium )

We find a strong synergism between tumor necrosis factor (TNF) and bacteria or their products. Endotoxin-"free" recombinant TNF, even at very high doses (160 ,ug), did not alone cause hemorrhagic necrosis (RN) in the skin of normal mice. Similarly, TNF alone had a low systemic toxicity in tumorand pathogen-free mice. However, TNF given intravenously with nanogram quantities of the endotoxin lipopolysaccharide caused lethal shock. Furthermore, subcutaneous injection of lipopolysaccharide made skin susceptible to subsequent induction ofHN by TNF injected in the same site 24 hr later. Mycoplasma-infected cells or corynebacteria also synergized with TNF to cause HN or lethal shock. In addition, we find that lymphotoxin, a cytokine functionally and genetically related to TNF, also synergized with the bacteria to cause HN, whereas interleukin la or interferon y did not. Together, the results indicate that a synergy between TNF and bacteria or their products causes HN and lethal shock in normal mice. Certain bacterial products, such as the endotoxin lipopolysaccharide (LPS), induce the release of a factor into serum that causes hemorrhagic necrosis (HN) of tumors and is called tumor necrosis factor (TNF; ref. 1). TNF has been cloned and the recombinant material, like the native product, causes HN of tumors (2). It is of major concern that, in addition, the recombinant TNF has been found to induce many of the deleterious effects of endotoxin, such as lethal shock and disseminated HN (3). It is not clear why this systemic toxicity of TNF in vivo was only found after recombinant preparations ofTNF became available and why it had not been observed in previous studies using nonrecombinant sources of TNF (for review, see ref. 4). The higher systemic toxicity of recombinant TNF could be due to the use of higher doses, to structural differences between the recombinant and nonrecombinant TNF molecules, or to impurities in the recombinant preparations (4). Interestingly, we find that recombinant TNF, which is virtually endotoxin-free, has a low toxicity in the absence of added endotoxin or other microbial agents so long as tumorfree and pathogen-free animals are tested. However, we observe a very strong synergism between TNF and LPS endotoxin or between TNF and other bacteria in causing lethal shock or HN in normal tissues. MATERIALS AND METHODS Mice. Sixto 10-week-old, pathogen-free, female mice of the C3H/HeN, C3H/HeJ, athymic NCR nu/nu, and BALB/cAn strains were purchased from a germ-freederived, defined-flora colony at the Frederick Cancer Research Institute (Frederick, MD) and maintained in pathogen-free conditions in laminar flow hoods. Cytokines and Bacteria. Recombinant human TNF (lot 3056-55) and recombinant human lymphotoxin (LT; lot 42601) had specific activities of 5.01 x 107 units (U)/mg and 1.02 x 108 U/mg, respectively, based on cytotoxicity of actinomycin D-treated L929 mouse fibroblast cells (5). Recombinant murine interferon y (IFN-,y; lot 4407-47) had a specific activity of 1.97 x 107 U/mg. TNF, LT, and IFN-y were obtained from Genentech. Recombinant human interleukin la (IL-la; lot SM63) had a specific activity of 5 x 109 U/mg and was obtained from Hoffmann-La Roche. Endotoxin levels of cytokines, expressed as endotoxin units (EU), were TNF c 0.125 EU/ml, IL-la < 15 EU/ml, IFN-y < 2 EU/ml, and LT c 0.062 EU/ml, as measured in the Limulus amoebocyte lysate assay given that 1 EU is equal to 0.1 ng/ml of USP standard Escherichia coli endotoxin (lot EC5). The LPS endotoxin of E. coli 0111:B4 (lot 743397, Difco) was diluted in nonpyrogenic saline (Invenex Labs, Melrose Park, IL) to 10 mg/ml and sterilized by 0.22-npm filtration before use. Heat-killed Corynebacterium parvum (lot 608122, Behring Diagnostics, San Diego, CA) was diluted in nonpyrogenic saline to 10 mg/ml before use. Mycoplasma orale was passaged by intracellular growth in vitro and expanded with the cells for the generation of lysate as described below. Cell Lines. The UV-induced fibrosarcomas 1591-RE and 1316-RE, normal heart-lung fibroblast strains, and 10T1⁄22 cells (an untransformed fibroblast line) were obtained and passaged as described (6). Assay for HN and Lethal Shock. The lower backs of mice were shaven and then depilated with a chemical depilator (Nair, Carter-Wallace, New York). Test material was injected subcutaneously (s.c.) with or without TNF. Removing the hair before the start of the experiment did not influence the results and for convenience, therefore, we usually depilated the skin before the start of the experiment. The degree ofHN was scored 24 hr later and in some cases skin sections were examined microscopically. Cell lysates used for induction of necrosis were generated by three cycles of freezethawing followed by a 30-min sonication. In some experiments, mice were primed for HN with LPS (100 ,tg) injected s.c. 24 hr before cytokine injection. To study the systemic toxicity of mixtures of TNF and LPS or TNF and C. parvum, the materials were coinjected i.v. by way of the retroorbital plexus.