Cutting Edge: Tissue Inhibitor of Metalloproteinase 3 Regulates TNF-Dependent Systemic Inflammation

More than any other cytokine TNF is central to the initiation and orchestration of inflammation. It directs circulating leukocytes to sites of infection by inducing selectin expression on endothelial cells and integrin expression on leukocytes. It promotes dendritic cell migration to lymph nodes and leukocyte migration to infected tissue via the induction of chemokine expression. It is involved in the formation of granulomas around unresolved infections. TNF also triggers clots in microvasculature, which is important for the containment of local infections and can promote cell growth, survival or death (1). Excessive or prolonged release of TNF underlies many human diseases. Although the pathophysiological importance of TNF is well illustrated by the success of anti-inflammatory therapies that rely on binding to and inactivating this cytokine (2), limitations in current therapies still remain. Understanding all aspects of TNF regulation is thus critical for developing novel therapies to control TNF activity.

TNF is expressed as a membrane-bound molecule and is released from the cell surface by proteolytic cleavage. Membrane-bound and cleaved TNF have distinct physiological effects as demonstrated in models of heart disease, arthritis, and systemic shock (3, 4, 5). Biochemically, p75/TNFRSF1B is more easily activated by membrane-bound TNF, whereas p55/TNFRSF1A readily responds to soluble TNF (5). The TNF receptors, like TNF itself, are subject to proteolytic cleavage and this process has also proved to be physiologically important. Prevention of TNFRSF1A proteolysis promotes liver inflammation and enhances sensitivity to septic shock (6). The major metalloproteinase responsible for TNF cleavage is a disintegrin and metalloprotease (ADAM) known as ADAM17 or TNF-α-converting enzyme (TACE) (7), which also cleaves both TNF receptors (8, 9). Additionally, TACE processes a number of other molecules involved in immunity and inflammation (9, 10, 11). TACE can swiftly alter the availability of TNF by cleaving it from myeloid and T cells, allowing the shed molecule to diffuse and act on the surrounding tissue, vasculature and at distant sites. Thus the regulation of TACE may be an important checkpoint for the magnitude of an inflammatory response.

Among the four tissue inhibitors of matrix metalloproteinases (TIMPs),² TIMP3 is the only one that binds to the extracellular matrix (12) and contains an amino acid sequence (PFG) required to inhibit TACE (13). TIMP3 is induced by molecules involved in inflammation, such as the proinflammatory agent PMA and the anti-inflammatory cytokine TGF-β. Aged timp3^−/− mice have a normal life span in C57BL/6 and FVB strains, but develop mild lymphocytic inflammation in the liver (14). In this study we investigate the importance of TIMP3 in the pathophysiology of systemic inflammation. LPS is a strong inducer of systemic shock, it triggers TNF release (15), which mediates LPS-induced shock (16). Challenging timp3-deficient mice with LPS revealed uncontrolled systemic inflammation leading to animal morbidity due to the dysregulated TNF ligand and receptor shedding. Despite concurrent increased shedding of TNF and its receptors, the net effect was elevated TNF signaling. These data highlight the importance of TIMP3 in regulating the inflammatory response.

Animal protocols were reviewed and approved by the animal care committee of the institute. Our previously generated timp3^−/− mice were backcrossed seven times or more into the FVB/N or C57BL/6 background (17). C57BL/6-Tnfrsf1a^tm1Mak animals were obtained from The Jackson Laboratory.

LPS was obtained from Sigma-Aldrich (serotype 0111:B4). Synthetic metalloproteinase inhibitor (AG3340; Pfizer) was prepared as described previously (18) and administered i.p. (2.5 mg/mouse) 90 min before LPS.

TNF, TNFRSF1B, and IL-6 ELISA kits were purchased from BD Pharmingen and TNFRSF1A ELISA reagents from R&D Systems.

Macrophages were derived from bone marrow following a standard protocol. Briefly, cells were plated overnight to remove stromal cells and mature resident macrophages. Nonadherent cells were transferred to ultra-low attachment polystyrene plates and differentiated with 50 ng/ml M-CSF over 5 days. A total of 2.5 × 10⁵ macrophages/well was allowed to adhere overnight and exposed to 0 or 100 ng/ml LPS for 8 h.

RNA expression of timp3 was quantified by TaqMan real-time RT-PCR using an ABI Prism 7700 sequence detection system as described elsewhere (19).

Statistical differences in TNF levels were calculated using the two-tailed Student’s t test and those in serum protein levels by two-way ANOVA. The area under the curve was calculated by the trapezoid rule and survival differences using the Fisher exact test.

We investigated whether the loss of TIMP3 affects the release of TNF using in vitro and in vivo systems. Macrophages are the major source of TNF following LPS induction. We established that wild-type primary bone marrow-derived macrophages (BMDM) express TIMP3 (Fig. 1,A). Macrophages obtained from timp3^−/− mice in response to LPS released significantly greater levels of TNF into the medium compared with those obtained from wild-type littermates. Eight hours after LPS exposure, soluble TNF levels were 80% higher in the medium of timp3^−/− macrophages (Fig. 1 B, p < 0.005). This suggests that macrophages can create their own TIMP3 to regulate TNF cleavage. TIMP3 is normally anchored to the extracellular matrix by binding to chondroitin sulfate and heparin sulfate (20). Monocytes have been shown to express chondroitin sulfate and heparan sulfate (21), potentially these proteoglycans allow TIMP3 to associate directly with the macrophage cell surface. Therefore, it is possible that macrophage-generated TIMP3 acts in concert with matrix-bound TIMP3 to inhibit TNF release even more powerfully in vivo.

To investigate the kinetics of TNF shedding in vivo, serum was collected from LPS-treated wild-type and timp3^−/− mice at 20-min intervals over 190 min after LPS injection (Fig. 2). Baseline levels of TNF were undetectable in both timp3^−/− and wild-type serum. In wild-type mice, TNF levels peaked sharply at 90 min, with an attenuated peak at 150 min. In contrast, TNF levels rose more rapidly in timp3^−/− mice and failed to return to the baseline levels during this time. Overall, the serum TNF levels remained elevated by ∼35% over the 3-h period, as determined by the area under the curve (Fig. 2 A). These data suggest that timp3 deficiency leads to accelerated TNF shedding and higher soluble TNF levels in response to LPS.

Biochemical studies show that in addition to cleaving TNF, TACE converts membrane-bound TNF receptors to their soluble forms (9). Both TNFRs are shed in response to LPS, and TNFRSF1A (p55) release is considered essential for normal down-regulation of the inflammatory response (6). Analysis of serum from LPS-stimulated mice revealed consistently elevated levels of both TNFRs over the course of 190 min following LPS injection in timp3-deficient mice (Fig. 2, B and C). The temporal pattern of soluble TNFRSF1B mirrored that of TNFRSF1A in timp3^−/− mice, showing similar dysregulation compared with wild-type mice. Thus, timp3 deficiency also affects the shedding of both TNF receptors.

Higher serum TNF increases TNF signaling, whereas greater shedding of TNF receptors can reduce TNF signaling by either binding soluble TNF or reducing the number of intact receptors available for this cytokine. Since timp3^−/− mice have increased shedding of both the TNF ligand and its receptors following LPS stimulation, we sought to determine the net outcome for TNF signaling. Serum IL-6 levels were used as a measure of TNF signaling, since the rise in IL-6 in response to LPS is partially TNF dependent (22, 23). We measured the levels of IL-6 in the serum from the same timed bleed experiment. IL-6 levels were substantially increased in timp3^−/− mice beginning at 70 min, with a 3-fold elevation by 150 min (Fig. 2 D), demonstrating an overall increase in LPS-stimulated TNF signaling in timp3-deficient mice.

LPS is a major component of Gram-negative bacteria and is a conventional trigger for inflammation mediated by an innate immune response. TNF is a principal mediator of the lethal effects of LPS (24). We therefore investigated the physiological importance of increased TNF signaling found in the timp3^−/− mice following LPS challenge. We subjected timp3 null mice and control littermates, both in the FVB background, to increasing LPS concentrations. At a sublethal dose of LPS, recovery of timp3^−/− mice was significantly delayed compared with control mice (5.4 days vs 4 days, p < 0.05), as assessed by weight loss (data not shown). timp3^−/− mice exposed to 200 μg of LPS showed significantly higher mortality than timp3^+/− control littermates. Only 20% of the timp3^−/− mice survived this dose compared with 80% or more of the control mice. This increased sensitivity to LPS was gender independent, similarly affecting male and female mice (Fig. 3, A and B). A dose of 600 μg of LPS was equally lethal for timp3 null and wild-type mice (Fig. 3 C). We also compared LPS sensitivity between wild-type and timp3^+/− mice and found timp3^+/− mice did not significantly differ in survival (data not shown), indicating no haploinsufficiency in this model of septic shock. Thus, the loss of TIMP3 leads to pathological inflammation due to an unregulated innate immune response.

One study has indicated that TIMP3 can exert an effect independent of its role as a metalloproteinase inhibitor (25). We asked whether TIMP3 functioned as a metalloproteinase inhibitor to control the inflammatory response following LPS challenge. A single dose of a broad-spectrum synthetic metalloproteinase inhibitor (MPi) (AG3340) was administered to mice before LPS injection. This treatment completely rescued the timp3^−/− mice from increased susceptibility to LPS, restoring the incidence of septic shock to levels found in control animals (Fig. 4,A). Furthermore, the increased serum TNF levels found in untreated timp3 null animals following LPS challenge dropped to those levels observed in control animals after timp3^−/− and wild-type mice were given AG3340 (Fig. 4 B). These data suggest that TIMP3 functions as a metalloproteinase inhibitor during the inflammatory response.

LPS-mediated septic shock is not always dependent on TNF signaling. For example, even in the absence of TNFRSF1A very high doses of LPS can cause septic shock (26). We asked whether increased susceptibility to septic shock in the absence of TIMP3 is dependent on TNF signaling, or occurs through a TNF-independent pathway. We generated double-deficient timp3^−/−/Tnfrsf1a^−/− mice in the C57BL/6 background. We first tested wild-type C57BL/6 for their response to LPS and found this strain to be generally more sensitive to LPS. Therefore, a lower dose of LPS (100 μg/mouse) was used for additional experiments. Ablation of Tnfrsf1a rescued the timp3^−/− mice from septic shock; specifically, timp3^−/−/Tnfrsf1a^−/− double-knockout mice were significantly less sensitive to 100 μg of LPS compared to the timp3^−/− mice (Fig. 4,C). As expected, Tnfrsf1a^−/− mice were completely resistant to LPS-induced shock at this dose. Furthermore, similar to the FVB mice, we found significantly higher serum TNF levels (Fig. 4,D) in the timp3^−/− C57BL/6 mice (▪) than those in their controls (□), taken 90 min after LPS injection. It has been previously reported that in response to LPS, serum TNF levels are significantly higher in Tnfrsf1a null mice compared to wild-type controls (27), suggesting this receptor is involved in lowering serum TNF levels. We also found that soluble TNF levels in the timp3^−/−/Tnfrsf1a^−/−double-deficient mice were significantly higher than those in controls (Fig. 4 D, black striped bars vs open bar). IL-6 levels for both timp3^+/ and timp3^−/− groups dropped when combined with Tnfrsf1a loss, 30% in timp3^+/ mice (p = 0.089) and 43% in timp3^−/− mice (p = 0.024). These data show that an intact TNF signaling pathway is necessary for the heightened LPS-induced inflammatory response in the timp3^−/− mice.

Inflammation is necessarily a rapid and powerful response that is essential for host survival yet has the potential for devastating consequences if not precisely controlled. timp3-deficient mice not only exhibit accelerated TNF shedding into serum, but the release of TNF receptors is also enhanced. Dysregulation of the TNF system results in a net increase in TNF signaling, reflected in IL-6 levels, leading to heightened septic shock and animal mortality. Thus, TIMP3 is a critical regulator of the release of presynthesized, membrane-bound mediators, dictating the magnitude of the inflammatory response. The source of TIMP3 regulating these events, stromal and/or hemopoietic, is an important question that remains to be investigated.

The innate immune system is an ancient response to infection, some elements of which are shared by all multicellular species (28). The TIMPs are also an ancient family of genes, strongly conserved throughout evolution (29). The loss of TIMP in Drosophila leads to autolysis and premature death in the adult fly (30). The Drosophila TIMP can inhibit human TACE; it is functionally and structurally most similar to TIMP3 (29). TIMP3 is unique among the mammalian TIMPs in its ability to inhibit TACE, the enzyme primarily responsible for releasing TNF (7, 31). In this capacity TIMP3 is ideally situated to protect the organism against an overactive innate immune response by inhibiting the wide release of a powerful initiator of inflammation. Our study was designed to understand an extracellular point of control in acute systemic inflammation, specifically focusing on TNF signaling, which plays a principal role in initiating inflammation. TNF regulation has become a focus for a new generation of anti-inflammatory therapies and we postulate that TIMP3-based therapies can be developed to control inflammation.

The authors have no financial conflict of interest.