Failure to Regulate TNF-Induced NF-κB and Cell Death Responses in A20-Deficient Mice

See allHide authors and affiliations

Science  29 Sep 2000:
Vol. 289, Issue 5488, pp. 2350-2354
DOI: 10.1126/science.289.5488.2350


A20 is a cytoplasmic zinc finger protein that inhibits nuclear factor κB (NF-κB) activity and tumor necrosis factor (TNF)–mediated programmed cell death (PCD). TNF dramatically increases A20 messenger RNA expression in all tissues. Mice deficient for A20 develop severe inflammation and cachexia, are hypersensitive to both lipopolysaccharide and TNF, and die prematurely. A20-deficient cells fail to terminate TNF-induced NF-κB responses. These cells are also more susceptible than control cells to undergo TNF-mediated PCD. Thus, A20 is critical for limiting inflammation by terminating TNF-induced NF-κB responses in vivo.

During inflammatory responses, TNF and interleukin-1 (IL-1) signals activate NF-κB, which regulates the transcription of other proinflammatory genes. The factors that limit these responses are poorly understood. A20 is a cytoplasmic protein thought to be expressed predominantly in lymphoid tissues, and heterologously expressed A20 can inhibit TNF-induced NF-κB and PCD responses in cell lines (1–4). A20 binding to TNF receptor–associated factor-2 (TRAF2), inhibitor of NF-κB kinase gamma (IKKγ), and/or A20-binding inhibitor of NF-κB activation (ABIN) suggest potential mechanisms by which A20 could regulate TNF receptor signals (5–7); however, the functions of A20 in vivo are unknown. Thus, we generated A20-deficient (A20−/−) mice by gene targeting (8).

A20+/− mice appeared normal without evidence of pathology. A20−/− mice were born from interbred A20+/−mice in Mendelian ratios, demonstrating that A20 is not required for embryonic survival. A20−/− pups were runted as early as 1 week of age and began to die shortly thereafter (Fig. 1A). Gross and histological examination of 3- to 6-week-old A20−/− mice revealed severe inflammation and tissue damage in multiple organs, including livers (Fig. 1, B and E), kidneys (Fig. 1, C and F), intestines (Fig. 1G), joints, and bone marrow (Fig. 1H). Flow cytometric analysis of A20−/− spleens and livers revealed increased numbers of activated lymphocytes (CD3+ CD44+), granulocytes (CD3 Gr-1+ Mac-1+), and macrophages (CD3 Mac-1+) (8). Double mutant A20−/− recombinase-activating gene-1–deficient (RAG-1−/−) mice developed granulocytic infiltration, cachexia, and premature death at a similar frequency and severity to A20−/− RAG-1+/− littermates (Fig. 1, F and J), indicating that lymphocytes are not required for the inflammation seen in A20−/− mice. Finally, skin sections revealed thickened epidermal and dermal layers without inflammation (Fig. 1I). Thus, A20 is essential for preventing spontaneous innate immune cell–mediated inflammation and tissue destruction, as well as regulating skin differentiation.

Figure 1

Generation and histology of A20−/− mice. (A) Gross appearance of 4-week-old A20+/+ and A20−/− mice. (B) Gross appearance of A20+/+ and A20–/– livers. Note pale acellular regions of A20−/− livers. (C) Gross appearance of A20+/+ and A20−/− kidneys. Note atrophied kidney in A20−/− mouse. (D) Flow cytometric analyses of TNF expression in A20+/+ and A20−/− splenocytes stimulated with LPS. (E) Hematoxylin- and eosin-stained (H and E) sections from A20+/+ and A20−/− livers. Note inflammation and hepatocyte loss in A20−/− livers. (F) H and E kidney sections. Note interstitial nephritis, glomerular dilatation, and cortical tubular atrophy in small A20−/−kidney. (G) H and E colonic sections. Note colitis, including lamina propria inflammation, crypt abscess, and branching epithelial crypts in A20−/− colon. (H) H and E joint and bone sections. Note bone marrow replacement with inflammatory cells, thinned trabecular bone, and destructive arthritis in A20−/− bone and joint. (I) H and E skin sections. Note thickened epidermis and dermis, and loss of hair follicles and fat in A20−/− skin. (J) H and E kidney sections from A20+/− RAG-1−/− (left panel) and A20−/− RAG-1−/− (right panel) mice. Note normal appearance of A20+/−RAG-1−/− kidney. Note interstitial nephritis, glomerular dilatation, and cortical tubular atrophy in A20−/−RAG-1−/− kidney (right panel), comparable to A20−/− RAG-1+/+ kidney [right panel in (F)].

The role of A20 in regulating inflammation was further evaluated by examining the sensitivity of A20−/− mice to lipopolysaccharide (LPS). All A20−/− mice died within 2 hours of injection of 5 mg LPS per kg of body weight, whereas A20+/+ and A20+/− mice given 5, 12, or 25 mg/kg LPS survived without significant morbidity (Table 1). This hypersensitivity to LPS was correlated with increased numbers of A20−/− splenocytes expressing TNF after LPS stimulation (Fig. 1D). In addition, A20−/− mice were highly susceptible to low doses of TNF, as all A20−/− mice died within 2 hours of injection of 0.1 mg/kg TNF, whereas A20+/+ and A20+/− mice given 0.1, 0.2, or 0.4 mg/kg TNF survived (Table 1). Consistent with the marked susceptibility of A20−/− mice to TNF, A20 mRNA expression was dramatically increased by TNF in all tissues examined from normal mice (Fig. 2A). Thus, A20 may protect mice from inflammatory mediators by regulating TNF responses in multiple cell types.

Figure 2

Sensitivity of A20−/− thymocytes and MEFs to TNF-mediated PCD. (A) Northern analysis of A20 mRNA expression in tissues from TNF-injected normal mice (LIV, liver; KID, kidney; SPL, spleen; THY, thymus; COL, colon; LN, lymph node). Comparable RNA loading and integrity was confirmed by ethidium staining of 28S rRNA. (B) Survival of A20−/−(solid bars) and A20+/− (hatched bars) thymocytes from 2- to 3-week-old mice 5 hours after in vitro treatment with the indicated agents. TNF was used at 10 ng/ml, and cycloheximide (CHX) was used at 10 μg/ml in all experiments. Asterisk indicatesP < 0.001 by Tukey's test. (C) Western analyses of IκBα [Santa Cruz Biotechnology (SCB)], phospho-SAPK/JNK [New England Biolabs (NEB)], SAPK/JNK (NEB), Bcl-x (Transduction Labs), and Bcl-2 (Pharmingen) proteins in lysates from TNF-treated thymocytes. (D) Northern analysis of A20 mRNA expression in TNF-treated A20+/+ MEFs. (E) Survival of A20−/− (solid bars) and A20+/−(hatched bars) MEFs 10 hours after in vitro treatment with the indicated agents. (F) Western analyses of TRAF2 (MedBiol Labs), cIAP-1 (Trevigen), phospho-SAPK/JNK, and SAPK/JNK proteins in lysates from TNF-treated MEFs.

Table 1

A20−/− mice succumb to sublethal doses of LPS and TNF. Indicated doses of LPS or TNF were given to 17- to 20-day-old A20+/+ and A20−/− littermates. Numbers of mice surviving at 2 hours are indicated over the numbers of mice injected. Surviving mice were observed for 6 hours further to rule out delayed effects of LPS or TNF.

View this table:

The hypersensitivity of A20−/− mice to TNF may be due in part to the capacity of A20 to regulate PCD (1,4). Thymocytes constitutively express both TNF (9) and A20 mRNA (4). Although corticosteroids, γ-irradiation, and Fas receptor ligation killed comparable numbers of A20+/− and A20−/−thymocytes, A20−/− thymocytes were more sensitive to TNF, both in the presence and absence of cycloheximide (Fig. 2B). TNF-mediated PCD was blocked by the caspase inhibitor ZVAD-fmk, confirming that caspase-dependent pathways kill these cells. Levels of the survival proteins Bcl-2 and Bcl-x were comparable in A20−/− and A20+/+ thymocytes (Fig. 2C). Both stress-activated protein kinase (SAPK) (or c-Jun N-terminal kinase, JNK) phosphorylation and inhibitor of κB alpha (IκBα) degradation were seen in TNF-treated A20−/− thymocytes (Fig. 2C), suggesting that the synthesis of survival proteins by SAPK/JNK- and NF-κB–dependent pathways was intact (10–13). Thus, A20 protects thymocytes from TNF-mediated PCD independently of protein synthesis or other known thymocyte survival factors.

The ability of A20 to regulate TNF responses was further examined in mouse embryonic fibroblasts (MEFs), which express negligible A20 mRNA at rest and dramatically increase levels of A20 mRNA expression after TNF treatment (Fig. 2D). While pretreatment of normal cells with TNF leads to the synthesis of survival proteins which protect these cells from subsequent TNF plus cycloheximide (14), A20−/− MEFs universally died despite TNF pretreatment (Fig. 2E). Activation of both SAPK/JNK and NF-κB pathways and similar levels of the survival proteins cellular inhibitor of apoptosis-1 (c-IAP1) and TRAF2 were seen in TNF-treated A20+/+ and A20−/− MEFs (Fig. 2F andFig. 3A). Thus, TNF-mediated synthesis of presumably all NF-κB– and SAPK/JNK-dependent survival proteins (15) except A20 was insufficient to protect A20−/− MEFs from PCD mediated by TNF plus cycloheximide.

Figure 3

Prolonged NF-κB responses to TNF in A20−/− MEFs. Electrophoretic mobility-shift assay (EMSA), Western, and Northern blot analyses of A20+/+and A20−/− MEFs treated repeatedly with TNF and harvested at the indicated time points. (A) EMSA analyses of NF-κB activity, using an NF-κB consensus oligonucleotide (SCB). (B) Western blot analysis of IκBα expression. (C) Northern blot analyses of IκBα and glyceraldehyde phosphate dehydrogenase (GAPDH) mRNA expression in MEFs. (D) Western blot analyses of IκBα and phospho-IκBα expression after proteasome inhibition. (E) IKK kinase assay of TNF-treated MEFs. Total cell lysates from repeatedly TNF-treated MEFs were immunoprecipitated with an anti-IKKγ antibody (SCB), and kinase activity was assessed using a GST-IκBα (1-54) substrate (upper panels). Comparable IKKβ protein in immunoprecipitated samples confirmed by Western analysis (lower panels). (F) Western analysis of IκBα expression in IL-1β–treated MEFs.

A20 inhibits NF-κB activation (2), and dysregulated NF-κB activity leads to inflammation and premature death in IκBα−/− mice (16). Moreover, the perturbed skin differentiation seen in A20−/− mice resembles the skin of IκBα−/− mice (16). Thus, the pathogenesis of A20−/−mice may be due in part to dysregulated NF-κB activity. Repeated TNF treatment of normal MEFs caused IκBα degradation and NF-κB binding to DNA, followed by down-regulation of NF-κB binding and reaccumulation of IκBα protein by 60 min (Fig. 3, A and B). In contrast, NF-κB binding to DNA persisted and IκBα protein was not detected in A20−/− MEFs from 60 to 180 min of TNF treatment (Fig. 3, A and B). IκBα mRNA levels, transcriptionally enhanced by NF-κB (17), increased in response to TNF in both A20+/+ and A20−/− MEFs, indicating that the failure of A20−/− MEFs to reaccumulate IκBα protein was not due to a failure to express IκBα mRNA (Fig. 3C). Addition of the proteasome inhibitor MG-132 to MEFs 15 min after TNF treatment caused A20−/− MEFs to regain normal levels of IκBα protein (Fig. 3D, top panels), suggesting that the lack of IκBα protein reaccumulation in TNF-treated A20−/− MEFs was due to rapid degradation of newly synthesized IκBα protein, rather than the failure of these cells to translate IκBα mRNA. IκBα protein that reaccumulated in MG-132–treated A20−/− but not A20+/+ MEFs was phosphorylated (Fig. 3D, bottom panels), suggesting that persistent IKK (a multimeric complex comprising IKKα, IKKβ, and IKKγ) activity caused rapid phosphorylation of newly synthesized IκBα protein in TNF-treated A20−/−MEFs. Direct measurement of IKK activity in lysates from TNF-treated MEFs confirmed this suggestion (Fig. 3E). Therefore, synthesis of IκBα mRNA and IκBα protein is insufficient to terminate NF-κB signals in the absence of A20.

Finally, we examined the role of A20 in regulating NF-κB responses to IL-1β. NF-κB activity increased and decreased normally and IκBα protein reaccumulated normally in IL-1β–treated A20−/− MEFs (Fig. 3F). Thus, although prior studies suggested that heterologous A20 can inhibit IL-1β–induced NF-κB responses (5, 18), A20 is not essential for terminating these responses. Moreover, it is likely that A20 inhibits TNF activation of the NF-κB pathway upstream of IKKγ, since IKKγ is required for both IL-1β– and TNF-induced NF-κB activation (19).

A20 is a dynamically regulated and pleiotropically expressed gene that is required for negatively regulating NF-κB responses in vivo. A20 may also regulate TNF-induced SAPK/JNK and PCD responses. The ability of A20 to inhibit TNF- but not IL-1β–induced NF-κB signals suggests these signals can be differentially regulated in vivo. The rapid expression of A20 is essential for limiting inflammatory responses and the damage those responses cause in multiple tissues.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: ama{at}


View Abstract

Stay Connected to Science

Navigate This Article