Proteolytic Inactivation of MAP-Kinase-Kinase by Anthrax Lethal Factor

See allHide authors and affiliations

Science  01 May 1998:
Vol. 280, Issue 5364, pp. 734-737
DOI: 10.1126/science.280.5364.734


Anthrax lethal toxin, produced by the bacterium Bacillus anthracis, is the major cause of death in animals infected with anthrax. One component of this toxin, lethal factor (LF), is suspected to be a metalloprotease, but no physiological substrates have been identified. Here it is shown that LF is a protease that cleaves the amino terminus of mitogen-activated protein kinase kinases 1 and 2 (MAPKK1 and MAPKK2) and that this cleavage inactivates MAPKK1 and inhibits the MAPK signal transduction pathway. The identification of a cleavage site for LF may facilitate the development of LF inhibitors.

Anthrax toxin, produced by the bacterium Bacillus anthracis, is composed of three proteins: protective antigen (PA), edema factor (EF), and lethal factor (LF) (1). PA binds to specific cell surface receptors and, upon proteolytic activation to a 63-kD fragment (PA63), forms a membrane channel that mediates entry of EF and LF into the cell (2). EF is an adenylate cyclase and together with PA forms a toxin referred to as edema toxin (3). LF and PA together form a toxin referred to as lethal toxin. Lethal toxin is the dominant virulence factor produced by B. anthracis and is the major cause of death of infected animals (4). Intravenous injection of lethal toxin into rats causes death in as little as 38 min (5), and addition of the toxin to mouse macrophages in culture causes lysis within 2 hours (6). LF is a 776–amino acid protein that contains a putative zinc-binding site [HEFGF (7)] at residues 686 through 690, which is characteristic of metalloproteases. Mutation of the H or E residues inactivates LF (8) and reduces its zinc-binding activity (9). However, no physiological substrate has been identified.

The National Cancer Institute maintains a database of antineoplastic drugs that have been tested against a panel of 60 human cancer cell lines [NCI's ADS (10)]. A screen of this database aimed at identifying novel inhibitors of the mitogen-activated protein kinase (MAPK) signal transduction pathway, an evolutionarily conserved pathway that controls cell proliferation and differentiation, revealed that anthrax LF had an activity profile similar to that of PD09859, a compound that selectively inhibits the MAPK pathway (11). We therefore examined the effect of LF on the MAPK pathway.

In response to extracellular signals, MAPK is phosphorylated and activated by MAPK kinases 1 and 2 (MAPKK1 and MAPKK2). In oocytes of the frog Xenopus laevis, progesterone-stimulated synthesis of Mos, a serine/threonine kinase, leads to activation of the MAPK pathway, which is essential for the activation of maturation-promoting factor (that is, cyclin B/p34cdc2 kinase) and the resumption of meiosis (maturation) (12). Addition of PA and LF to oocyte culture medium had no effect on progesterone-induced oocyte maturation (13). In contrast, injection of 1 ng of LF into oocytes inhibited maturation by 50% as judged by an assay of germinal vesicle (nuclear envelope) breakdown (GVBD), and GVBD was completely inhibited by 10 ng of LF (Table 1). Injection of LF Glu687 → Cys687 (E687C), an inactive LF containing a single amino acid substitution in the putative zinc-binding site (8), had no effect on GVBD (Table 1). Because a decrease in adenosine 3′,5′-monophosphate–dependent protein kinase A activity is also required for oocyte maturation (12), there was concern that low levels of EF may have been present as a contaminant. However, preparations of LF from strains ofB. anthracis deficient in the production of EF also blocked oocyte maturation (Table 1). In contrast, LF did not inhibit GVBD induced by injection of Δ90 cyclin B, a truncated nondegradable form of cyclin B and a potent activator of p34cdc2 kinase (14) (Table 1).

Table 1

Effects of anthrax LF on Xenopus oocyte maturation. Oocytes were isolated, defolliculated, injected, and induced to mature with progesterone as described (22). After progesterone-treated control oocytes had completed GVBD, oocytes were fixed and dissected to score GVBD. In experiment 1, LF was purified from culture supernatants of B. anthracis Sterne, a strain that produces PA, LF, and EF, with the use of methods described previously (23), and GVBD was scored by dissection 20 hours after progesterone treatment. In experiment 2, LF was produced as a recombinant fusion protein, PA20-LF, and cleaved with Factor X to release the PA20 domain (8). The expression host was B. anthracis lacking the EF gene. GVBD was scored 4 hours after progesterone treatment, when GVBD was complete in control oocytes. The injection buffer was 0.1 M KCl and 10 mM Hepes (pH 7.5).

View this table:

The addition of LF, but not of LF E687C, inhibited Mos-induced activation of MAPK in oocyte lysates (Fig.1A), which suggests that oocytes cannot mature in the presence of LF because of a failure in MAPK activation. When immunoblots of these lysates were probed with antibodies to the COOH-terminus of MAPKK1, we detected antigen throughout the incubation with a slightly increased mobility (albeit at reduced levels in comparison to control lysates). In contrast, antibodies to the NH2-terminus of MAPKK1 did not detect the antigen after the addition of LF. These results suggest that LF may have proteolytically modified MAPKK1, rendering it undetectable by antibodies against its NH2-terminus.

Figure 1

LF inhibition of MAPK phosphorylation. (A) LF or LF E687C (4 μg from a stock containing 1 mg/ml) was added to 40 μl of oocyte lysate (24), which was activated 0.5 hour later by the addition of 2.6 μg of maltose-binding protein–Mos fusion protein [stock (0.75 mg/ml) purified from bacteria (25)]. Samples were taken at 1-hour intervals; analyzed by SDS-PAGE (22) with antibodies against phosphorylated MAPK (PO4-MAPK, New England Biolaboratories, 1:1000), MAPK (Zymed clone ERK-7D8, 1:1000), or the COOH- or NH2-terminus of MAPPK1 [MAPKK1 (CT) or (NT); Upstate Biotechnology; 1:500]; and visualized by chemiluminescence. (B) NIH 3T3 cells expressing the V12-S35 Hrasoncogene were grown to approximately 70% confluence in Dulbecco's minimal essential medium (DMEM) + 10% fetal bovine serum (FBS) and then incubated in DMEM + 10% FBS containing PA (1 μg/ml) for 10 min. Control medium (C), LF, or LF E687C (E) (0.1 μg/ml) was then added directly to the cells. Cells were lysed in lysis buffer [20 mM Pipes (pH 7.4), 150 mM NaCl, 1 mM EGTA, 1.5 mM MgCl2, 1% SDS, aprotonin and leupeptin (10 μg/ml), 1 mM phenylmethylsulfonyl fluoride, and 1 mM sodium orthovanadate] at the times indicated and clarified by centrifugation (15,000g for 15 min at 4°C). Samples (10 μg) were analyzed as described above, except that antibodies against PO4-MAPK were obtained from Promega (1:15,000).

We tested the effects of LF on tumor-derived NIH 3T3 (490) cells that express a form of the human V12HRas oncogene (V12-S35 Hras) that has a mutant effector domain. This mutant of Ras constitutively activates the MAPK pathway but is defective in other effector functions (15). The addition of PA and LF, but not LF E687C, to these cells inhibited MAPK activation (Fig. 1B). This inhibition was accompanied by an increase in the electrophoretic mobility of MAPKK1 observed with the COOH-terminal antibody, as well as a loss of MAPKK1 epitopes observed with the NH2-terminal antibody, observations that are consistent with the notion that LF proteolytically modifies MAPKK1.

We next ascertained the effects of LF on MAPK activation in vitro by assaying myelin basic protein (MBP) phosphorylation in the presence of MAPKK1 and MAPK. The addition of LF, but not LF E687C, prevented MBP phosphorylation (Fig.2A). To exclude the possibility that contaminants in the LF preparation inhibited MBP phosphorylation, we adsorbed LF to the proteolytically activated PA heptamer (PA63), to which it tightly binds (16), and re-purified it by column chromatography. In all cases, the activity that inhibited MBP phosphorylation coeluted with LF (Fig. 2B).

Figure 2

LF inhibition of in vitro MAPK activation. (A) His6-MAPKK1 prepared from bacterial lysates (17) (0.25 μg from a stock containing 0.1 mg/ml) was incubated for 15 min at 30°C in the presence or absence of 0.25 μg of LF or LF E687C and then assayed for MAPK activity (MBP phosphorylation) (26). (B) Comigration of MAPKK1 inhibitory activity with LF repurified by adsorption to PA63. LF was chromatographed on a MonoQ HR5/5 column in the presence or absence of the PA63 heptamer. The MonoQ column was eluted with a gradient of NaCl in 10 mM 2-(N-cyclohexylamino)ethane sulfonic acid (CHES) and 0.06% aminoethanol (pH 9.0). The samples applied to the columns were 250 μg of LF, 250 μg of PA63, and 250 μg of LF + 350 μg of PA63. Fractions were pooled and assayed for inhibition of MAPKK1 activity (26). The numbered peaks correspond to the lane position in the autoradiograph.

The increase in electrophoretic mobility of MAPKK1 and the disappearance of NH2-terminal epitopes inXenopus oocyte lysates and ras-transformed cells (Fig. 1) suggested that LF might proteolytically cleave MAPKK1. We tested this hypothesis by adding LF to a His6-tagged MAPKK1 fusion protein produced in bacteria. Within seconds of LF addition, the apparent molecular mass of MAPKK1 decreased by ∼6 to 8 kD (Fig.3A). Cleavage by LF was enzymatic, because proteolysis of MAPKK1 was observed within 15 min with as little as 2 ng of LF per 200 ng of MAPKK1 (≅1 mol of LF: 400 mol of MAPKK1)(Fig. 3B). LF also increased the electrophoretic mobility of His6-tagged MAPKK2 (Fig. 3B).

Figure 3

Proteolysis of MAPKK1 by LF. (A) His6-MAPKK1 (0.1 μg) was incubated in 16 μl of assay buffer (26) in the presence of 1 μg of LF E687C or 1 μg of LF. Samples were withdrawn at 0, 10, or 20 min and analyzed by SDS-PAGE and protein immunoblotting with antibodies against the COOH-terminus of MAPKK1. (B) His6-MAPKK1 or His6-MAPKK2 (0.2 μg) was incubated in the presence of 2 μg of LF E687C or LF that had been serially diluted in ADB. Aliquots were withdrawn at 15 and 30 min and analyzed by SDS-PAGE and immunoblotting with antibodies against the COOH-terminus of MAPKK1 or the NH2- and COOH-terminus of MAPKK2 (Upstate Biotechnology; 1:1000). (C) His-tagged MAPKK1 deletion mutants (0.1 μg) isolated from bacterial lysates (17) were incubated in assay buffer (26) in the presence or absence of LF (1 μg) for 15 min at 30°C and analyzed by protein immunoblotting. Δn1 and Δn2 were completely resistant to proteolysis, whereas Δn3, Δn4, and Δn6 were cleaved. Δn5 showed partial resistance to proteolysis, which suggests that structural modifications in this construct may partially hinder LF activity.

Because the MAPKK1 used in these analyses was His6-tagged at the NH2-terminus, the actual decrease in its mass was 5 kD less (17), which suggests that LF cleaves MAPKK1 in the first 30 amino acids. To test this directly, we assayed NH2-terminal deletion mutants of MAPKK1 (17) for their ability to serve as substrates for LF. These analyses showed that Δn3 (32–51), Δn4 (44–51), Δn5 (38–43), and Δn6 (32–37) were susceptible to LF proteolysis, whereas Δn1 (1–32) and Δn2 (1–52) were resistant (Fig. 3C). Thus, the NH2-terminal 32 amino acids are essential for cleavage or binding of MAPKK1 by LF. To determine the exact site of cleavage, we performed NH2-terminal sequence analysis of the larger MAPKK1 proteolytic fragment (18) and identified the amino acid sequence IQLNPAPDG (7), which corresponds to amino acids 8 through 16 of MAPKK1. Thus, LF cleaves MAPKK1 between amino acids 7 and 8, resulting in the loss of the NH2-terminal seven residues [PKKKPTP (7)]. These results also suggest that these NH2-terminal residues are essential for MAPKK1 activity. Consistent with this, bacterially produced MAPKK1 lacking the NH2-terminal seven residues was resistant to proteolysis by LF and possessed no activity toward MAPK (Fig.4, A and B). These results agree with previous findings that MAPKK1 mutants with deletions of the NH2-terminal 32 amino acids are less active than wild-type kinase (19). In addition, the NH2-terminal 32 amino acids of MAPKK1 contain a MAPK binding site (20), which suggests that LF may prevent the association of MAPKK1 with its substrate.

Figure 4

Analysis of the cleavage site in MAPKK1. (A) His6-tagged MAPKK1 as well as His6-tagged MAPKK1 mutants lacking the seven NH2-terminal residues (ΔN7) (27) were treated with LF E687C or LF and assayed for cleavage by protein immunoblotting as described (Fig. 3). (B) MAPKK1 and Δn7 were assayed for MAPK activity (MBP phosphorylation) in the presence of LF E687C or LF as described (Fig. 2A). (C) His6-tagged MAPKK1 (0.1 μg) as well as His6-tagged MAPKK1 mutants (0.1 μg) containing proline-to-alanine mutations at residues 5 (P5A) or 7 (P7A) were incubated for 15 min in the presence of 1 μg of LF E687C or LF that had been serially diluted in ADB. Samples were analyzed by SDS-PAGE and immunoblotting with antibodies against the COOH-terminus of MAPKK1.

Our results indicate that frog, mouse, and human MAPKK1 are all substrates of LF. In addition, NH2-terminal sequence analysis of the larger MAPKK2 proteolytic fragment (18) revealed that LF cleaved MAPKK2 between residues 9 and 10, resulting in the loss of NH2-terminal residues LARRKPVLP (7). MAPKK1 and MAPKK2 share a similar NH2-terminal sequence consisting of three positively charged residues followed by two prolines that are separated by one or two amino acids. Because other residues near the cleavage site are not conserved, the basic residues and prolines may constitute the sequence recognized by LF. This is consistent with a recent report that LF cleaves synthetic peptides after the sequence RRP (7,21). We generated MAPKK1 mutants in which either Pro5 or Pro7 was changed to Ala and assayed each mutant for its ability to serve as a substrate for LF. These analyses showed that compared to wild-type MAPKK1, both mutants were resistant to LF cleavage (Fig. 4C), indicating that the proline residues constitute an important component of the cleavage site.

Further characterization of the cleavage site for LF will be an important step in the development of LF inhibitors. Although LF activity correlates with the activity of the MAPKK inhibitor PD09859 in the NCI's ADS, we cannot exclude the possibility that LF has other cellular substrates. Indeed, MAPKK3 (but not MAPKK4) contains a similar sequence at its NH2-terminus, and sequences resembling the putative consensus cleavage site occur frequently in protein databases. However, because the MAPK signaling pathway plays such a fundamental role in signal transduction, it is likely that inhibition of MAPKK activities is important in the pathogenesis of anthrax.

  • * Present address: Biopraxis, Post Office Box 9100–78, San Diego, CA 92191, USA.


Stay Connected to Science

Navigate This Article