Cutting Edge: TNF-α-Converting Enzyme (TACE/ADAM17) Inactivation in Mouse Myeloid Cells Prevents Lethality from Endotoxin Shock¹

The proinflammatory cytokine TNF-α has potent beneficial and detrimental functions (1). Its dysregulation triggers autoimmune diseases such as rheumatoid arthritis and Crohn’s disease, but TNF is also critical for host defense against pathogens such as listeria. TNF is synthesized as a membrane-anchored precursor that is released by the TNF-α-converting enzyme (TACE)³ a disintegrin and metalloprotease (ADAM) 17 (ADAM17) (2, 3). However, other enzymes, including ADAM10 (4), ADAM19 (5), matrix metalloproteinase 7 (6), and proteinase 3 (7) have also been implicated in releasing TNF, raising questions about the identity of the relevant TNF convertase in primary cells in vivo. LPS-triggered endotoxin shock in mice, a model for septic shock, provides an excellent means to evaluate the mechanism and consequences of TNF processing in vivo (8, 9, 10). LPS stimulates TNF release via TLRs and the resulting endotoxin shock requires soluble, but not membrane-anchored, TNF (8) released from myeloid cells (9) and can be prevented with metalloprotease inhibitors (10). This has led to the hypothesis that TACE is the critical enzyme for endotoxin-triggered release of TNF from myeloid cells in vivo in mice. However, to date this hypothesis could not be tested because TACE-deficient mice die shortly after birth (11, 12).

To address the role of TACE in endotoxin shock in vivo, we generated conditional TACE-deficient mice. We found that the temporal systemic deletion of Tace by the inducible Mx1-Cre transgene (13) or its deletion in myeloid cells by the M lysozyme promoter (LysM-Cre) (14) protected the mice from endotoxin shock with serum TNF levels that were significantly lower than those in control animals. These observations corroborate the hypothesis that endotoxin shock in mice is caused by TACE-dependent shedding of TNF from myeloid cells.

Genomic 129/SVJ DNA was used to generate a 1.6-kb PCR product with a noncoding sequence 1.1 kb upstream of exon 2 and a 5.2-kb PCR product with exon 2 and the surrounding noncoding region, which were subcloned into pCR2.1 TOPO vectors (Invitrogen Life Technologies). A loxP and an EcoRI site were inserted 1.0 kb downstream of exon 2, and the long and short arms were subcloned into a pKOII vector (provided by Dr. R. DePinho, Dana-Farber Cancer Center, Boston, MA). 129P2/OlaHsd cell clones harboring floxed Tace were identified by Southern blot and PCR analysis (Fig. 1, A and B). Chimeric mice were generated by C57BL/6 blastocyst injection and bred with C57BL/6 mice to generate floxed Tace heterozygous mice (referred to as Tace^flox/+; formal allele nomenclature: Adam17^tm1Bbl). Homozygous Tace^flox/flox mice were viable and fertile and did not display any evident pathological phenotype or histopathological defects (data not shown). Tace^flox/+ mice were mated with Tg(EIIa-cre)C5379Lmgd transgenic mice (15) (referred to as EIIa-Cre) to generate Tace-null heterozygous mice (referred to as Tace^+/−; formal allele nomenclature: Adam17^tm1.1Bbl) or with Lyzs^tm1(cre)Ifo (referred to as LysM-Cre) or Tg(Mx1-cre)1Cgn (referred to as Mx1-Cre) transgenic mice (13, 14) to generate conditional TACE-deficient mice. For temporal deletion of floxed Tace, 6-wk-old Tace^flox/flox/Mx1-Cre⁺ mice were injected i.p. with 250 μg of polyinosinic-polycytidylic acid (pIpC; Sigma-Aldrich) three times at 2-day intervals as described (13). In all experiments with Tace^flox/flox/Cre⁺ mice, littermate Tace^flox/flox/Cre⁻ mice served as controls.

Tissues were fixed in 4% paraformaldehyde and PBS, sectioned, stained with H&E, and photographed with a DXM1200 camera (Nikon) on a BX50 microscope (Olympus). Images were processed with Adobe Photoshop CS2.

Embryonic fibroblasts isolated from E13.5 Tace^−/− embryos as described (16) were immortalized with the SV40 large T Ag. Absence of TACE protein in Tace^−/− fibroblasts was confirmed by Western blotting (see Fig. 1 C).

Immortalized fibroblasts were transfected with alkaline phosphatase-tagged TNF or TGFα (see Ref. 16 for details). Fresh Opti-MEM (Invitrogen Life Technologies) with or without 25 ng/ml PMA was added for 1 h after 1 day, and alkaline phosphatase activity was measured by colorimetry or visualized by in-gel staining of substrates as described (16).

Bone marrow was collected from the tibiae and femurs of 6-wk-old Tace^flox/flox/LysM-Cre⁻ or littermate Tace^flox/flox/LysM-Cre⁺ mice. RBCs were removed with RBC lysis buffer (Roche), and the remaining cells were plated on tissue culture plates. Adherent cells were grown in DMEM with 10% FCS, antibiotics, and 50 ng/ml recombinant mouse macrophage CSF (WAKO) for 4 days and then used as bone marrow macrophages and incubated with or without 1 μg/ml LPS (Sigma-Aldrich) for 3 h. Soluble TNF was measured by ELISA (R&D Systems).

Endotoxin shock was induced by i.p. injection of 5 μg of LPS (Sigma-Aldrich) and 20 mg of d-galactosamine (WAKO). All injected mice were closely monitored every hour for the first 16 h and every 3–6 h thereafter. To enhance serum TNF levels in a separate experiment (see Fig. 3 C), Tace^flox/flox/LysM-Cre⁻ or littermate Tace^flox/flox/LysM-Cre⁺ mice were treated with 100 μg of LPS and 20 mg of d-galactosamine and sacrificed after 3 h. Serum TNF was measured by ELISA. All animal experiments were approved by the Institutional Animal Care and Use Committee of the Keio University School of Medicine (Tokyo, Japan).

A Student’s t test for two samples assuming equal variances was used to calculate the p values. p < 0.05 was considered statistically significant.

Mice containing a floxed Tace allele were crossed with a germline “Cre-deleter” strain (EIIa-Cre) (15) to generate Tace^+/− animals that were then bred to produce Tace^−/− mice (Fig. 1,A; see Materials and Methods for details). Removal of floxed Tace was confirmed by PCR and Southern blotting (Fig. 1,B) and the lack of TACE protein was corroborated by Western blotting (Fig. 1,C). The Tace^−/− animals generated in this study were indistinguishable from the previously described Tace^ΔZn/ΔZn mice that lack the exon carrying the catalytic site of TACE (2, 11); they died shortly after birth (Table I), with open eyes and defects in the aortic, pulmonic, and tricuspid heart valves (Fig. 2, A and B; pulmonic valves not shown). In addition, constitutive and phorbol ester-stimulated shedding of transfected TGF-α and TNF was strongly reduced in Tace^−/− mouse embryonic fibroblasts (mEF) compared with wild-type controls (Fig. 2 C) as previously described for Tace^ΔZn/ΔZn mEF (5, 11, 16). Thus, Cre-mediated excision of floxed Tace results in the inactivation of TACE.

To evaluate the role of TACE in adult mice, we generated Tace^flox/flox/Mx1-Cre animals to allow excision of the floxed exon by i.p. injection of pIpC (13) in 6-wk-old mice. The pIpC-induced recombination in Mx1-Cre mice occurs in various organs with different efficiency, with almost complete recombination in bone marrow, liver, and spleen (13). Accordingly, we found strongly decreased amounts of floxed TACE in these tissues in Tace^flox/flox/Mx1-Cre⁺ mice 1 wk after the pIpC injection (Fig. 3,A). After treatment with pIpC, Tace^flox/flox/Mx1-Cr⁺ and Tace^flox/flox/Mx1-Cre⁻ mice displayed no evident pathological phenotypes or histopathological defects for at least 2 wk (data not shown). We next tested the survival of Tace^flox/flox/Mx1-Cre⁺ animals in a murine model for endotoxin shock, which depends on soluble TNF in vivo (8). When Tace^flox/flox/Mx1-Cre⁺ mice or Tace^flox/flox/Mx1-Cre⁻ controls were injected i.p. with LPS and d-galactosamine, six of eight Mx1-Cre⁺ mice were protected from endotoxin shock whereas seven of eight Mx1-Cre⁻ controls died within the first 16 h after injection (Fig. 3,B). Serum TNF levels after LPS injection were also significantly lower in Tace^flox/flox/Mx-Cre⁺ mice (230 pg/ml, ± 177 SD; n = 7) compared with wild-type controls (852 pg/ml, ± 349 SD; n = 8) or Tace^flox/flox/Cre⁻ animals (779 pg/ml, +/− 310 SD; n = 10) (Fig. 3 C). These results support the hypothesis that the shedding of TNF by TACE is crucial for LPS-induced endotoxin shock in mice, although protection from endotoxin shock could also be due, at least in part, to the reduction in cleavage of other TACE substrates.

To further narrow down the cell type in which TACE is required for endotoxin-induced release of TNF and the resulting lethality, we generated Tace^flox/flox/LysM-Cre⁺ mice to remove TACE in myeloid cells (14), which are critical for LPS-induced shock (9). Tace^flox/flox/LysM-Cre⁺ mice were viable and fertile and did not display any evident abnormalities (data not shown). Macrophages isolated from Tace^flox/flox/LysM-Cre⁺ mice had very low amounts of the floxed Tace allele, as confirmed by PCR, and of the TACE protein, as detected by Western blot analysis, compared with the Tace^flox/flox/LysM-Cre⁻ controls (Fig. 3,D). LPS-stimulated release of soluble TNF was strongly reduced in macrophages isolated from Tace^flox/flox/LysM-Cre⁺ animals compared with Tace^flox/flox/LysM-Cre⁻ controls (Fig. 3,E), which is consistent with the recently reported ablation of TNF shedding from Tace^ΔZn/ΔZn leukocytes isolated from a radiation chimera (17). When Tace^flox/flox/LysM-Cre⁺ animals and Tace^flox/flox/LysM-Cre ⁻controls were subjected to LPS and d-galactosamine to trigger endotoxin shock, 11 of 13 control animals perished within 8 h of injection whereas only three of 13 Tace^flox/flox/LysM-Cre⁺ animals succumbed to this challenge (Fig. 3,F). Following endotoxin challenge, the average TNFserum levels in Tace^flox/flox/LysM-Cre⁺ (212 pg/ml, ± 158 SD; n = 7) were very similar to the levels in Tace^flox/flox/Mx1-Cre⁺ mice (230 pg/ml, ± 177 SD; n = 7) and, thus, significantly lower than in Tace^flox/flox/Cre⁻ animals or wild-type controls (see above and Fig. 3,C). The remaining endotoxin-stimulated shedding of TNF in Tace^flox/flox/LysM-Cre⁺ and Tace^flox/flox/Mx-Cre⁺ is most likely due to cells in which one or both floxed Tace alleles escaped Cre excision (see Fig. 3, A and D), although it cannot be ruled out that other candidate TNF sheddases also contribute to TNF shedding following Cre excision of TACE.

Taken together, these results provide strong evidence that TACE is indeed a principal enzyme responsible for the release of soluble TNF from the relevant primary cells (myeloid cells) during endotoxin shock in an intact organism in vivo (a discussion of criteria to link enzymes and relevant substrates can be found in Ref. 18). These results are consistent with studies of mice lacking an endogenous inhibitor of TACE, the tissue inhibitor of matrix metalloproteinases 3 (TIMP3) (19), that have an exacerbated response to an endotoxin challenge (20). In addition to releasing TNF, TACE is known to have a role in the shedding of numerous other membrane proteins from mEFs (11, 16, 21), and TACE activation in leukocytes also triggers shedding of the TNF receptors I and II (17). Shedding of these and other membrane anchored cytokines and growth factors could thus potentially contribute to the anergic state of the immune system at later stages of septic shock. It would therefore be interesting to determine whether or not TACE inhibitors might offer advantages in the treatment of septic shock in humans over the blocking of TNF, at least in those cases where an involvement of TNF is likely (reviewed in (22). Moreover, our results further support the notion that TACE is likely also a good target for treatment of other pathologies caused by soluble TNF, most notably rheumatoid arthritis, Crohn’s disease, and psoriasis. The Tace^flox/flox animals described here will help address these questions, and permit an evaluation of other TACE functions during mouse development and in disease models, including those related to the role of TACE in epidermal growth factor receptor-ligand activation in cancer (23).

We thank Dr. Ronald A. DePinho, Dana-Farber Cancer Institute (Boston, MA) for the pKOII vector, Dr. Irmgard Förster, University of Düsseldorf (Düsseldorf, Germany) for the LysM-Cre transgenic mouse line, Dr. Ursula Lichtenberg, University of Cologne (Cologne, Germany) for the Mx1-Cre transgenic mouse line, Dr. Chingwen Yang from the Rockefeller University transgenic facility (New York, NY) for ES-cell targeting, Dr. Willie Mark, Sloan-Kettering Institute (New York, NY) for blastocyst injection and production of chimeric mice, Ms. Yuko Hashimoto and Ms. Shizue Tomita for technical assistance, and Dr. Roy Black for critically reading this manuscript.

The authors have no financial conflict of interest.