Efficient growth inhibition of Bacillus anthracis by knocking out the ribonucleotide reductase tyrosyl radical Eduard Torrents*, Margareta Sahlin*, Daniele Biglino†‡, Astrid Gra¨slund†, and Britt-Marie Sjo¨berg*§ Departments of *Molecular Biology and Functional Genomics and †Biochemistry and Biophysics, Arrhenius Laboratories for Natural Sciences, Stockholm University, SE-106 91 Stockholm, Sweden Edited by Harry B. Gray, California Institute of Technology, Pasadena, CA, and approved October 13, 2005 (received for review July 27, 2005)
Bacillus anthracis, the causative agent of anthrax, is a worldwide problem because of the need for effective treatment of respiratory infections shortly after exposure. One potential key enzyme of B. anthracis to be targeted by antiproliferative drugs is ribonucleotide reductase. It provides deoxyribonucleotides for DNA synthesis needed for spore germination and growth of the pathogen. We have cloned, purified, and characterized the tyrosyl radical-carrying NrdF component of B. anthracis class Ib ribonucleotide reductase. Its EPR spectrum points to a hitherto unknown three-dimensional geometry of the radical side chain with a 60° rotational angle of C␣-(C-C1)-plane of the aromatic ring. The unusual relaxation behavior of the radical signal and its apparent lack of line broadening at room temperature suggest a weak interaction with the nearby diiron site and the presence of a water molecule plausibly bridging the phenolic oxygen of the radical to a ligand of the diiron site. We show that B. anthracis cells are surprisingly resistant to the radical scavenger hydroxyurea in current use as an antiproliferative drug, even though its NrdF radical is efficiently scavenged in vitro. Importantly, the antioxidants hydroxylamine and N-methyl hydroxylamine scavenge the radical several orders of magnitude faster and prevent B. anthracis growth at several hundred-fold lower concentrations compared with hydroxyurea. Phylogenetically, the B. anthracis NrdF protein clusters together with NrdFs from the pathogens Bacillus cereus, Bacillus thuringiensis, Staphylococcus aureus, and Staphylococcus epidermidis. We suggest the potential use of N-hydroxylamines in combination therapies against infections by B. anthracis and closely related pathogens. anthrax 兩 electron paramagnetic resonance 兩 hydroxyurea 兩 N-methyl hydroxylamine 兩 bacterial growth inhibition
B
acillus anthracis is a zoonotic soil organism and the causative agent of anthrax. This bacterium is a highly virulent mammalian pathogen and continues to be a worldwide problem in domesticated and wild animals in Africa and Asia. Infection from inhalation of B. anthracis spores can result in a high rate of mortality (1). For humans, this bacterium has been the recent focus of attention when after September 11, 2001 it was used as a biological weapon by delivery of B. anthracis spores to several locations through the U.S. Postal Service. These events represent the first reported bioterrorism-related outbreaks of anthrax (2). Existing treatment for inhalation anthrax involves administration of antibiotics shortly after exposure and several months of long use of ciprofloxacin or doxycycline with potentially unpleasant side effects (3). Of importance to prevent the disease is the prompt use of antimicrobial prophylaxis after a suspected bioterrorism attack. The search for additional potential antiproliferative targets in B. anthracis is of great importance. Ribonucleotide reductase (RNR) is one such target. RNR is an essential enzyme in all living organisms. It catalyzes the production of deoxyribonucleotides for DNA synthesis by reduction of their corresponding ribonucleotides via intricate radical chemistry (4). Even though all known RNRs appear to have a common evolutionary relationship, contemporary RNRs can be grouped into three major classes (I, II, and III) based 17946 –17951 兩 PNAS 兩 December 13, 2005 兩 vol. 102 兩 no. 50
on differences in the mode of radical generation and in overall quaternary structure (4). Whereas almost all eukaryotic organisms only encode class I RNRs, prokaryotic organisms are known to quite often encode more than one class of RNR principally depending on environmental life styles (5). Class I RNRs are tetrameric (␣22) enzymes, composed of one homodimeric protein (␣2) that includes the active site and another homodimeric protein (2) that carries a stable free radical at a tyrosine residue close to a diiron-oxo center. Generation of the essential tyrosyl radical requires oxygen, and class I RNRs are functional only during aerobiosis. Class I RNRs are further subdivided into class Ia and Ib, whose major differences relate to the fine-tuning of their allosteric control executed by binding of nucleoside triphosphates to the ␣2 protein (6). The class Ia genes nrdA and nrdB code for the ␣2 protein R1 (or NrdA) and the 2 protein R2 (or NrdB). Most class Ib operons contain the nrdE and nrdF genes [for the ␣2 protein NrdE (or R1E) and the 2 protein NrdF (or R2F), respectively] accompanied by the nrdI gene encoding a flavodoxin-like protein of unknown function, plus or minus an nrdH gene encoding a thioredoxin-like physiological reductant for class Ib RNR. Class II RNRs are coenzyme B12-dependent and can work under both aerobiosis and anaerobiosis. Class III RNRs carry a stable but oxygensensitive glycyl radical and are functional only during anaerobiosis. The release of the complete genome sequence from the aerobic, Gram-positive, sporulating B. anthracis Ames strain (7) and a number of other B. anthracis strains兾isolates allowed the identification of genes coding for class Ib and III RNRs in this genome. The class Ib genes are assembled in the order nrdI, nrdE, and nrdF, as in many other microorganisms (8), and are plausibly forming an operon. The nrdH is found elsewhere on the B. anthracis chromosome. Interestingly, the B. anthracis nrdE gene deviates from other bacterial genes and includes a phage-like group I intron that interrupts the gene in two exons (7). The only RNR capable of maintaining aerobic growth in B. anthracis is class Ib RNR, because class III RNR is restricted to anaerobic conditions. There is relatively little knowledge about the mechanism of DNA synthesis in B. anthracis. The essential tyrosyl radical of the class Ib NrdF protein is a possible target for tailored antibacterial drugs based on radical scavenging. In this article, we characterize the unique biochemical properties of B. anthracis NrdF and show that its tyrosyl radical as well as growth of the pathogen is extremely sensitive to the antioxidants hydroxylamine (HA) and N-methyl Conflict of interest statement: No conflicts declared. This paper was submitted directly (Track II) to the PNAS office. Abbreviations: DTT, dithiothreitol; HA, hydroxylamine; HU, hydroxyurea; M-HA, N-methyl hydroxylamine; RNR, ribonucleotide reductase. Data deposition: The sequence reported in this paper has been deposited in the GenBank database (accession no. AJ704809). ‡Present address: Max Planck Institute for Bioinorganic Chemistry, D-45470 Mu ¨ lheim an der
Ruhr, Germany. §To whom correspondence should be addressed. E-mail:
[email protected].
© 2005 by The National Academy of Sciences of the USA
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0506410102
Fig. 1. Sequence alignment of NrdFs from B. anthracis (Ba), C. ammoniagenes (Ca), and S. typhimurium (St), NrdF2 from M. tuberculosis (Mt), and R2兾NrdB from E. coli (Ec). The numbering on the right and left refers to protein sequence positions, and the middle numbers refer to the number of residues between the displayed sequence regions. Gray background, conserved residues; black background with asterisk, tyrosyl radical harboring residue; black background, iron ligands (liganded iron ion according to ref. 4). GenBank accession numbers are as follows: BaNrdF, AJ704809; CaNrdF, CAA70766 (Y09572); StNrdF, NP㛭461734; MtNrdF2, NP㛭216497; EcNrdB, NP㛭416738.
Materials and Methods Bacterial Strains, Plasmids, and Chemicals. Wild-type B. anthracis
Sterne 7700 pXO1⫺兾pXO2⫺, obtained from the Swedish Defence Research Agency, was used for PCR amplification of genomic DNA. Escherichia coli DH5␣F⬘ (Stratagene) and Rosetta (DE3) (Novagen), and plasmids pGEM-T easy from Promega and pET28a and pET22b from Novagen were used for recombinant DNA techniques. Potential RNR inhibitors were hydroxyurea (HU) and HA from Sigma; p-methoxyphenol, resveratrol, and methoxyamine from ICN; and M-HA, hydroxyguanidine, 1-methyl-1-hydroxyurea, 3-methyl-1-hydroxyurea, and 3,4,5-trihydroxybenzohydroxamic acid (all kind gifts from L. Thelander, Umeå University, Umeå, Sweden). Survival Experiments. For determining survival of B. anthracis and
E. coli B0 wild-type strains in the presence of selected potential inhibitors, cells were grown in LB medium to mid log phase (A550 ⬇ 0.5). Aliquots of 100 l from dilution series were plated on LB plates containing freshly prepared potential inhibitor at the indicated concentrations (see Fig. 5). Cloning of B. anthracis nrdF in Expression Vectors. The nrdF coding sequence from chromosomal DNA of B. anthracis Sterne 7700 was cloned into pET28a and pET22b, as described in Supporting Text, which is published as supporting information on the PNAS web site. The resulting plasmids pETS138 and pETS145, respectively, were transformed into E. coli Rosetta (DE3) generating the ETS101 strain for production of His-tagged NrdF protein and the ETS102 strain for production of non-His-tagged NrdF protein (see Supporting Text). Reconstitution of B. anthracis NrdF with Iron. In vitro incorporation of iron was performed as described in ref. 9. For reconstitution directly in the cuvette, we added 2–4 eq of Fe2⫹ [usually 3 l of 5 mM (NH4)2Fe(SO4)2 in 0.1 M HCl] to 500 l of 10 M apoprotein in 50 mM Tris䡠HCl (pH 7.6), and the spectrum was registered after ⬇15 sec. For B. anthracis NrdF, the protein concentration was determined by the Bradford assay (Bio-Rad), using BSA as standard, and for E. coli and mammalian R2兾NrdB, extinction coefficients (280–310) of 120,000 and 124,000 M⫺1䡠cm⫺1, respectively, were used. Iron analyses were as described by Fish (10). UV-Vis Absorption Spectroscopy and Kinetics of Incubations with Radical Scavengers. A PerkinElmer Lambda 2 spectrophotometer
was used for scanning spectra and kinetic experiments recorded at 25°C. Each potential inhibitor substance was freshly dissolved in concentrations to enable administrations of 2–5 l to start the reaction. Typically, kinetic scans were started ⬇10 sec after addition of the drug. Kinetic traces were recorded at 409 nm for B. anthracis NrdF, at 410 nm for E. coli R2兾NrdB, and at 415 nm for mammalian R2兾NrdB. Reconstitution with 3 eq of Fe2⫹ per B. anthracis NrdF Torrents et al.
dimer immediately preceded drug incubation. The E. coli R2兾NrdB did not require reconstitution, and samples of mouse R2兾NrdB had been reconstituted previously according to the anaerobic procedure (11). Starting spectra were recorded before addition of the drug and after each kinetic trace. Subtraction of end-point spectrum from starting spectrum was used to evaluate the amount of radical [ ⫽ 3,250 M⫺1䡠cm⫺1 (12)] in each experiment. Electron Paramagnetic Resonance Measurements. EPR spectra were
recorded with a Bruker eleXsys 580 X-band spectrometer, equipped with an Oxford ESR900 cryostat for temperatures from 10 to 50 K and a liquid nitrogen flow system for temperatures above 90 K. The double integral of the spectrum was compared with that of a frozen copper standard solution to estimate the radical concentration in the sample. First-derivative EPR spectra were recorded at different microwave powers (P) and various temperatures to determine the microwave power at half saturation (P1/2) for each temperature. The experimental values were fit with the function I ⫽ 1兾(1 ⫹ (P兾P1/2))b/2, where I denotes the intensity of the EPR signal, and b is a component relating to the type of relaxation; the b factor is 1 for a completely inhomogeneous relaxation and 3 for a completely homogeneous relaxation. The EPR parameters of the spectra were extracted with the iterative simulation program (13). The main features of the spectra could be simulated assuming a radical with an unpaired electron coupled to 3 protons. The least square method iterative program (13) calculated the values for the g-tensor and for the hyperfine tensors (hf-tensor) of the three protons. The simulated spectrum with the best fitting parameters was computed with the program Bruker WINEPR SIMFONIA V.1.25. Phylogenetic Analyses. Sequence alignments were done with
(14) with the default gap opening and extension penalties. The neighbor-joining algorithm was used to recover protein distance matrix in CLUSTAL X and excluding positions with gaps. CLUSTAL X V.1.8.1
Results Cloning, Expression, and Purification of B. anthracis NrdF. Using the
sequence information of the B. anthracis Ames strain, we designed primers to amplify, clone, and sequence the nrdF gene (GenBank accession no. AJ704809) from the B. anthracis Sterne 7700兾 pXO1⫺兾pXO2⫺ strain. Also recently, the complete genomic sequence of B. anthracis Sterne 7700 has become available (NrdF GenBank accession no. YP㛭027539). All hitherto sequenced B. anthracis strains have identical NrdF sequences and ⬇40% amino acid sequence identity to the NrdF proteins in Fig. 1. Comparison of the B. anthracis primary structure to the crystallized E. coli R2兾NrdB (15, 16) and Salmonella typhimurium, Corynebacterium ammoniagenes, and Mycobacterium tuberculosis NrdF proteins (17– 19) revealed that all of the six important amino acids that act as ligands for the -oxo-bridged diiron site (4) are conserved in B. anthracis NrdF (Fig. 1). The potential radical carrying tyrosine is conserved (Fig. 1), and all residues postulated to participate in the PNAS 兩 December 13, 2005 兩 vol. 102 兩 no. 50 兩 17947
BIOCHEMISTRY
hydroxylamine (M-HA). Our results may form the basis for using N-hydroxylamines in combination therapies against respiratory infections of B. anthracis.
Fig. 2. UV-visible spectrum of iron-reconstituted B. anthracis NrdF protein. A 10 M solution of B. anthracis NrdF protein was reconstituted in the cuvette by addition of 3 eq of Fe2⫹, and the spectrum was recorded after 10 min (solid line). Arrows, iron charge transfer bands and the tyrosyl radical. Dashed line, HU-treated sample. Dotted line, subtraction of the HU-treated spectrum from the iron-reconstituted spectrum; the ⌬A409 corresponds to 4.25 M tyrosyl radical (12). (Inset) SDS兾PAGE (PhastGel gradient 10 –15, Amersham Biosciences) of 3-g samples of purified NrdF (lane 3) and His-tagged NrdF (lane 2) from B. anthracis.
radical transfer between NrdE and NrdF are present (data not shown). We used the pET system to overexpress in E. coli the B. anthracis NrdF protein and study its biophysical properties. The final yield of purified proteins was ⬇9 mg of NrdF per liter of culture in LB medium. SDS兾PAGE showed high purity of the purified proteins with estimated molecular masses of ⬇45 kDa (Fig. 2 Inset). The predicted molecular mass for this protein is 37 kDa, and a similar discrepancy with the experimental observed size was previously observed with the C. ammoniagenes NrdF protein (20). The B. anthracis NrdF protein had an estimated purity of 95% for His-tagged NrdF and 90% for non-His-tagged NrdF. The proteins contained no iron (apoprotein) and displayed no EPR signal (Fig. 3b). The Tyrosyl Radical of B. anthracis NrdF Has Unique Properties. After in vitro reconstitution of the apoprotein with Fe2⫹兾O2 (9), the NrdF contained 3.7 eq of Fe per dimer. The UV-visible spectrum of the reconstituted protein displayed a sharp band at 409 nm, indicative of the tyrosyl radical, and broader bands at 325 and 365 nm, indicative of a diiron-oxo center (Fig. 2, solid line). Treatment of the reconstituted protein with the radical scavenger HU gives the spectrum shown in Fig. 2 (dashed line), and subtraction of the two spectra (Fig. 2, dotted line) displays all of the typical features of a tyrosyl radical spectrum with the sharp 409-nm band and a broader 390-nm band. It was found that 3 eq of Fe per NrdF dimer gave optimal reconstitution regarding tyrosyl-radical concentration (0.4–0.5 radicals per NrdF dimer) and a spectrum with distinct iron charge transfer bands (data not shown). The radical is stable for 2–3 h at 25°C in dilute solution (10 M). We have used EPR to probe the B. anthracis radical cofactor in overproducing whole cells (data not shown), early purification steps (Fig. 3a), purified protein (Fig. 3b), and reconstituted protein (Fig. 3c). The cells can form the radical兾diiron site, but it is lost from the protein upon purification. In this respect, B. anthracis NrdF is more similar to mouse R2兾NrdB that also purifies as apoprotein (11) than to E. coli R2兾NrdB, S. typhimurium NrdF, and M. tuberculosis NrdF2 that retain their radicals upon purification. As expected, the 17948 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0506410102
Fig. 3. EPR spectra of NrdF from B. anthracis. (a) Crude extract at 100 K (microwave power 2 mW, modulation amplitude 0.2 mT). (b) Purified protein (11.8 mg兾ml) at 10 K (microwave power 2 mW, modulation amplitude 0.05 mT). (c) Purified reconstituted protein (13.4 mg兾ml) at 20 K (recording parameters as in b). (d) Simulation of spectrum c. (e) Purified reconstituted protein at room temperature (microwave power 20 mW, modulation amplitude 0.1 mT).
B. anthracis NrdF spectrum in Fig. 3c has typical characteristics of a tyrosyl radical. Its EPR signal is a hyperfine (hf ) doublet of ⬇1.5 mT splitting centered at g ⫽ 2.0055 and with additional smaller splitting of ⬇0.7 mT. The latter arise from hf coupling to the two identical C3- and C5-protons of the aromatic ring and are diagnostic for tyrosyl radicals (see Scheme 1, which is published as supporting information on the PNAS web site) (21). Additional hf couplings arise from the two nonequivalent C-protons of the tyrosyl radical, but the coupling of the second  proton (H2) is too small to be resolved in the spectrum. The splittings depend on their overlap with the spin at C1 of the aromatic ring and will vary with rotation along the C1-C bond. The rotation angle is defined as the angle C␣-(C-C1), the plane of the aromatic ring (see Scheme 2, which is published as supporting information on the PNAS web site). The observed spectrum of B. anthracis NrdF (Fig. 3) differs from the spectra of all class Ia and Ib R2 proteins (NrdB兾NrdF) studied so far. We have used simulations to identify the couplings and found the best fit with AH1x ⫽ 1.68, AH1y ⫽ 1.48, AH1z ⫽ 1.55, AH2x ⫽ 0.20, AH2y,z ⫽ 0.02, AH3,5x ⫽ ⫺0.91, AH3,5y ⫽ ⫺0.48, and AH3,5z ⫽ ⫺0.79 mT and with a line width tensor LWx ⫽ 0.64 mT, LWy ⫽ 0.63 mT, and LWz ⫽ 0.48 mT. The simulated g-values were fitted to gx ⫽ 2.0100, gy ⫽ 2.0042, and gz ⫽ 2.0026. The gx-value is considerably higher than any other value observed for a tyrosyl radical (see Table 3, which is published as supporting information on the PNAS web site); gx-values higher than 2.008 are considered diagnostics of radicals with a non-hydrogen bonded phenoxy group (22–24). Also Torrents et al.
Table 1. H hyperfine couplings and corresponding rotation angle for tyrosyl radicals in class I RNRs Average values of Ax, Ay, and Az, mT
Class Ib B. anthracis S. typhimurium M. tuberculosis Class Ia Mouse E. coli T4
H 1
H2
, °
Ref.
1.57 0.91 0.84
0.08 ⬍0.3 ⬍0.5
60 75 82
This work 28 29
2.06 2.06 1.96
0.59 ⫺0.03 0.75
35 30 20
24 26 25
AH1 of the B. anthracis radical differs from other RNRs and is smaller than for class Ia but larger than for previously characterized class Ib radicals (Table 1) (23–29). The observed spectrum can be the result of a different rotational angle in the B. anthracis radical compared with other RNR tyrosyl radicals. We have use the hf couplings of the  protons as described in ref. 30 to calculate for the new radical to 60° (Table 1). Progressive microwave saturation curves give information about the relaxation properties of the observed radical and in RNR R2 proteins (NrdB兾NrdF) also give an indication of the distance between the radical and the fast-relaxing diiron site. Evaluated parameters are P1/2 and the inhomogeneity parameter b (see Material and Methods). The P1/2 values for B. anthracis NrdF and their temperature dependencies coincide with those of M. tuberculosis NrdF2 but differ from those of mouse and E. coli R2兾NrdB radicals in the temperature region between 10 and 100 K (Fig. 4). The low P1/2 values for B. anthracis NrdF indicate that the interaction between the radical and the diiron site is weak, either because of a longer distance between them and兾or weak magnetic properties of the diiron site. The increase of the slope above 100 K can result from a change in relaxation mechanism of the diiron site or some movement in the active site. The change is consistent with the observation that the best fit with b-values are ⬍1 at temperatures below 100 K (Fig. 4), indicating a dipolar interaction (31) with the diiron site, but increases to ⱖ1 above 100 K when other relaxation
Fig. 5. Effects of potential RNR inhibitors on the growth of E. coli Bo and B. anthracis Sterne 7700. Percentage of survival of E. coli (open symbols) and B. anthracis (filled symbols) cells in the presence of HU (circles), HA (squares), M-HA (diamonds), or p-methoxyphenol (triangles). (Inset) Drug concentrations up to 0.5 mM for better resolution of potent inhibitor results. Each point represents the mean plating efficiency (⬇10 –1,000 colonies) from three consecutive 10-fold dilutions.
processes take over. In addition, the similarity of the B. anthracis EPR spectra at 10 K (Fig. 3c) and at room temperature (Fig. 3e) indicates a weak interaction between the radical and the diiron center, and that the structure around the tyrosyl radical is quite rigid even at room temperature. Similar behavior has been observed for the M. tuberculosis NrdF2 (29). B. anthracis Cells Are Extremely Sensitive to the Radical Scavengers HA and M-HA. The possibility to inhibit DNA synthesis via RNR
might be a good alternative or complement to the use of different antibiotics. We therefore compared the growth behavior of B. anthracis toward a set of well known radical scavengers and used as reference the growth behavior of E. coli under the same growth conditions. As seen in Fig. 5, B. anthracis Sterne 7700 is extremely resistant to HU [LD50 ⫽ 59 mM compared with an E. coli wildtype strain (LD50 ⫽ 3 mM)]. B. anthracis is also more resistant to p-methoxyphenol (LD50 ⫽ 2.4 mM) compared with E. coli (LD50 ⫽ 0.09 mM). In contrast, B. anthracis is more sensitive to HA (LD50 ⫽ 0.06 mM) and M-HA (LD50 ⫽ 0.17 mM) compared with E. coli (LD50 of 0.3 mM for each drug). Hence, B. anthracis is several orders of magnitude more sensitive to HA and M-HA as compared with HU. The Tyrosyl Radical of B. anthracis NrdF Is More Sensitive to HA and M-HA than Is That of the Mammalian R2兾NrdB. To study whether the
Fig. 4. P1/2 dependence on temperature for different tyrosyl radicals. Depicted are B. anthracis NrdF (䊐) (this work), M. tuberculosis NrdF2 (Œ) (29), E. coli R2兾NrdB (F) (this work and ref. 40), mouse R2兾NrdB (}) (40), and UV-induced tyrosyl radical in borate buffer (ƒ) (40).
Torrents et al.
growth inhibitions could be ascribed to scavenging of the RNR tyrosyl radical, we used real-time monitoring of disappearance of the sharp electronic absorption band of the tyrosyl radical at ⬇410 nm. As reference for B. anthracis NrdF, we used mouse R2兾NrdB and E. coli R2兾NrdB. The radicals in B. anthracis, mouse, and E. coli RNRs have approximately the same sensitivity toward HU (Table 2). Of great interest to the focus of this study was that HA and M-HA scavenged the B. anthracis tyrosyl radical ⬇250- and ⬇160fold more efficiently than did HU, whereas the mouse radical was not scavenged at all by M-HA (Table 2). Based on a detection limit of ⬇2 ⫻ 10⫺4 sec⫺1 for k1 in our assays, we can estimate that the B. anthracis radical is more than two orders of magnitude more sensitive toward M-HA compared with the mouse radical. Other PNAS 兩 December 13, 2005 兩 vol. 102 兩 no. 50 兩 17949
BIOCHEMISTRY
Origin of protein species
Table 2. HU, HA, and M-HA sensitivity of the tyrosyl radicals in B. anthracis NrdF, E. coli R2兾NrdB, and mouse R2兾NrdB R2 protein B. anthracis NrdF Mouse R2兾NrdB E. coli R2兾NrdB
HU (10 mM)* k1, sec⫺1
HA (1 mM)† k1, sec⫺1
M-HA (1 mM)† k1, sec⫺1
0.0022 ⫾ 0.0001 0.0039 ⫾ 0.0001 0.0035 ⫾ 0.0001
0.056 ⫾ 0.002 0.0095 ⫾ 0.003‡ 0.0027 ⫾ 0.0002
0.035 ⫾ 0.001 No radical decay§ 0.0005 ⫾ 0.00001
*Standard deviation from regression analyses. †Mean error from at least three independent decay analyses. ‡Overestimated decay rate, because HA causes concomitant loss of iron site; tyrosyl radical not completely lost even at long incubation times or in presence of 10 mM HA. §No radical decay seen for 10 min, i.e., detection limit for k ⬍ ⫺1 1 ⬍ 0.0002 s .
known radical scavengers were much less efficient inhibitors of B. anthracis NrdF (1-methyl-1-hydroxyurea, 3-methyl-1-hydroxyurea, and hydroxyguanidine), did not scavenge at all (methoxyamine, resveratrol, and 3,4,5-trihydroxybenzohydroxamic acid), or were equally efficient inhibitors of both B. anthracis NrdF and mouse R2兾NrdB (p-methoxyphenol) (data not shown). Discussion More than 60% of bacteria carrying a class Ib RNR depend on this class of RNR for growth during aerobiosis, either because their genomes only encode class Ib RNR or, as in the case of B. anthracis, their genomes encode a class Ib RNR in combination with a class III RNR that is only functional during anaerobiosis. The vast majority of Gram-positive bacteria rely on class Ib for normal growth, and, interestingly, many pathogens belong to this group. Our initial observation that B. anthracis relies on class Ib RNR for aerobic growth, whereas the mammalian host organism uses class Ia RNR, prompted a thorough characterization of the B. anthracis class Ib RNR. In this article, we have characterized the biochemical properties of the class Ib-specific NrdF component that carries a
stable tyrosyl radical necessary for class Ib enzyme activity in B. anthracis. The EPR spectrum of B. anthracis NrdF shows a signal that is similar to but distinctly different from previously reported tyrosyl radicals in proteins. The main differences in hf-tensor have been found in the -methylene protons, reflecting different conformations of the locked tyrosyl ring relative to the backbone of the protein. The conformation can be described by the rotational angle (Scheme 2) related to the dihedral angles for the two methylene protons. For class Ia radicals, is typically 20–35°, and for class Ib it is typically 75–85° (Table 1). According to the amino acid sequence and the suggested operon structure, the B. anthracis NrdF protein clearly belongs to the class Ib family; but from the hf-tensor of the H1, we could evaluate a angle of 60°, distinctly different from the typical class Ib geometry and suggesting a somewhat different structure at the iron兾radical site. The high gx-value of the simulated spectrum suggests a non-hydrogen bonded tyrosyl radical. In addition, the different relaxation behavior of the B. anthracis (and M. tuberculosis) NrdF radical compared with the mouse and E. coli R2兾NrdB radicals suggests a weaker exchange interaction
Fig. 6. Unrooted phylogenetic tree of representative NrdF proteins. In squares are shown the rotational angle for B. anthracis NrdF (this work), M. tuberculosis NrdF2 (29), and S. typhimurium NrdF (28). Only bootstrap values below 950 are shown. All sequences were from the National Center for Biotechnology Information database. 17950 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0506410102
Torrents et al.
We thank Åke Forsberg and Ulla Eriksson (Swedish Defence Research Agency, NCB Protection Division, Umeå, Sweden) for providing the B. anthracis Sterne 7700 strain, and MariAnn Westman, Torbjo ¨rn Astlind, and Britta Tumlin for excellent technical assistance. This work was supported by the Swedish Cancer Foundation, the Swedish Research Council, the Carl Trygger Foundation, and the Magn. Bergvall Foundation. E.T. was supported by a postdoctoral fellowship from the Ministerio de Educacio ´n y Ciencia (Spain).
1. Mock, M. & Fouet, A. (2001) Annu. Rev. Microbiol. 55, 647–671. 2. Jernigan, D. B., Raghunathan, P. L., Bell, B. P., Brechner, R., Bresnitz, E. A., Butler, J. C., Cetron, M., Cohen, M., Doyle, T., Fischer, M., et al. (2002) Emerg. Infect. Dis. 8, 1019–1028. 3 . Inglesby, T. V., O’Toole, T., Henderson, D. A., Bartlett, J. G., Ascher, M. S., Eitzen, E., Friedlander, A. M., Gerberding, J., Hauer, J., Hughes, J., et al. (2002) J. Am. Med. Assoc. 287, 2236–2252. 4. Eklund, H., Uhlin, U., Fa¨rnegårdh, M., Logan, D. T. & Nordlund, P. (2001) Prog. Biophys. Mol. Biol. 77, 177–268. 5. Torrents, E., Aloy, P., Gibert, I. & Rodriguez-Trelles, F. (2002) J. Mol. Evol. 55, 138–152. 6. Reichard, P. (2002) Arch. Biochem. Biophys. 397, 149–155. 7. Read, T. D., Peterson, S. N., Tourasse, N., Baillie, L. W., Paulsen, I. T., Nelson, K. E., Tettelin, H., Fouts, D. E., Eisen, J. A., Gill, S. R., et al. (2003) Nature 423, 81–86. 8. Torrents, E., Roca, I. & Gibert, I. (2003) Microbiology 149, 1011–1020. 9. Huque, Y., Fieschi, F., Torrents, E., Gibert, I., Eliasson, R., Reichard, P., Sahlin, M. & Sjo ¨berg, B.-M. (2000) J. Biol. Chem. 275, 25365–25371. 10. Fish, W. W. (1988) Methods Enzymol. 158, 357–364. 11. Mann, G. J., Gra¨slund, A., Ochiai, E., Ingemarson, R. & Thelander, L. (1991) Biochemistry 30, 1939–1947. 12. Petersson, L., Gra¨slund, A., Ehrenberg, A., Sjo ¨berg, B.-M. & Reichard, P. (1980) J. Biol. Chem. 255, 6706–6712. 13. Bernhard, W. A. & Fouse, G. W. (1989) J. Magn. Reson. 82, 156–162. 14. Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. (1997) Nucleic Acids Res. 25, 4876–4882. 15. Nordlund, P., Sjo ¨berg, B.-M. & Eklund, H. (1990) Nature 345, 593–598. 16. Ho ¨gbom, M., Galander, M., Andersson, M., Kolberg, M., Hofbauer, W., Lassmann, G., Nordlund, P. & Lendzian, F. (2003) Proc. Natl. Acad. Sci. USA 100, 3209–3214. 17. Uppsten, M., Davis, J., Rubin, H. & Uhlin, U. (2004) FEBS Lett. 569, 117–122. 18. Eriksson, M., Jordan, A. & Eklund, H. (1998) Biochemistry 37, 13359–13369. 19. Ho ¨gbom, M., Huque, Y., Sjo ¨berg, B.-M. & Nordlund, P. (2002) Biochemistry 41, 1381–1389. 20. Fieschi, F., Torrents, E., Toulokhonova, L., Jordan, A., Hellman, U., Barbe´, J., Gibert, I., Karlsson, M. & Sjo ¨berg, B.-M. (1998) J. Biol. Chem. 273, 4329–4337. 21. Lendzian, F. (2005) Biochim. Biophys. Acta 1707, 67–90.
22. Un, S., Atta, M., Fontecave, M. & Rutherford, A. W. (1995) J. Am. Chem. Soc. 117, 10713–10719. 23. Gerfen, G. J., Bellew, B. F., Un, S., Bollinger, J. M., Stubbe, J., Griffin, R. G. & Singel, D. J. (1993) J. Am. Chem. Soc. 115, 6420–6421. 24. Schmidt, P. P., Andersson, K. K., Barra, A. L., Thelander, L. & Gra¨slund, A. (1996) J. Biol. Chem. 271, 23615–23618. 25. Sahlin, M., Gra¨slund, A., Ehrenberg, A. & Sjo ¨berg, B.-M. (1982) J. Biol. Chem. 257, 366–369. 26. Bender, C. J., Sahlin, M., Babcock, G. T., Barry, B. A., Chandrashekar, T. K., Salowe, S. P., Stubbe, J., Lindstro ¨m, B., Petersson, L., Ehrenberg, A. & Sjo ¨berg, B.-M. (1989) J. Am. Chem. Soc. 111, 8076–8083. 27. Hoganson, C. W., Sahlin, M., Sjo ¨berg, B.-M. & Babcock, G. T. (1996) J. Am. Chem. Soc. 118, 4672–4679. 28. Allard, P., Barra, A. L., Andersson, K. K., Schmidt, P. P., Atta, M. & Gra¨slund, A. (1996) J. Am. Chem. Soc. 118, 895–896. 29. Liu, A., Po ¨tsch, S., Davydov, A., Barra, A. L., Rubin, H. & Gra¨slund, A. (1998) Biochemistry 37, 16369–16377. 30. Himo, F., Gra¨slund, A. & Eriksson, L. A. (1997) Biophys. J. 72, 1556–1567. 31. Galli, C., Innes, J. B., Hirsh, D. J. & Brudvig, G. W. (1996) J. Magn. Reson. B 110, 284–287. 32. Strand, K. R., Karlsen, S., Kolberg, M., Rohr, A. K., Gorbitz, C. H. & Andersson, K. K. (2004) J. Biol. Chem. 279, 46794–46801. 33. de Melo, E. J., Mayerhoffer, R. O. & de Souza, W. (2000) FEMS Microbiol. Lett. 185, 79–82. 34. Lori, F., Kelly, L. M., Foli, A. & Lisziewicz, J. (2004) Expert Opin. Drug Safety 3, 279–288. 35. Kaubisch, A., Kaleya, R., Haynes, H., Rozenblit, A. & Wadler, S. (2004) Cancer Chemother. Pharmacol. 53, 337–340. 36. Gale, G. R., Kendall, S. M., McLain, H. H. & Dubois, S. (1964) Cancer Res. 24, 1012–1020. 37. Dawes, S. S., Warner, D. F., Tsenova, L., Timm, J., McKinney, J. D., Kaplan, G., Rubin, H. & Mizrahi, V. (2003) Infect. Immun. 71, 6124–6131. 38. Atamna, H., Paler-Martinez, A. & Ames, B. N. (2000) J. Biol. Chem. 275, 6741–6748. 39. Kalns, J., Morris, J., Eggers, J. & Kiel, J. (2002) Biochem. Biophys. Res. Commun. 297, 506–509. 40. Sahlin, M., Petersson, L., Gra¨slund, A., Ehrenberg, A., Sjo ¨berg, B.-M. & Thelander, L. (1987) Biochemistry 26, 5541–5548.
Torrents et al.
PNAS 兩 December 13, 2005 兩 vol. 102 兩 no. 50 兩 17951
BIOCHEMISTRY
to the exceptional properties of the cell wall envelope of this bacterium, but this has not been the focus of our study. Here, we instead show that the radical scavengers HA and M-HA can penetrate the cell envelope and inhibit the growth of B. anthracis very efficiently (LD50 of 0.06 and 0.17 mM, respectively). Biochemical studies confirmed that the B. anthracis tyrosyl radical is a prime target for these drugs and showed that the mammalian radical was considerably more resistant toward HA and not inhibited at all by M-HA under these conditions. Interestingly, M-HA and several other N-substituted hydroxylamines have been found to delay senescence in human lung fibroblasts and senescence-accelerated mice (38). Anthrax, caused by B. anthracis infection, is one of the most feared diseases in the world today. Treatments need to be introduced almost before symptoms are seen, and the therapy needs to be fast-acting, especially in respiratory anthrax, where rapid initiation of therapy is essential for host survival (3). However, the delayed effect of antibiotic treatment limits its use to a narrow time window after exposure to lethal anthrax infection (39), and therefore combination therapy with drugs that work on a more shortterm basis is desirable. This is particularly important if engineered highly virulent and兾or multidrug-resistant strains are used as offensive weapons that confound prophylaxis or treatment. Discovery of key B. anthracis proteins that can be targeted by new drugs or vaccines is consequently an important task. After infection due to purposely delivered or naturally spread spores, development of the anthrax disease requires DNA replication for spore germination and for growth of the pathogen. In this respect, RNR is a promising drug target because of its key role in providing precursors for the DNA synthesis, which essentially prevents resistance development toward drugs of the radical scavenger type. Our results clearly demonstrate the possible use of M-HA for inhibiting the growth of B. anthracis.
with the diiron site and兾or a dipolar interaction with excited states of the antiferromagnetic site. Our observations suggest a longer distance between the B. anthracis tyrosyl radical and the iron site, possibly with a water molecule between the phenolic oxygen and an iron ligand. The distance from the tyrosyl oxygen to the closest iron in S. typhimurium NrdF is 6.7 Å (18), in mouse R2兾NrdB is 5.9 Å (32), and in E. coli R2兾NrdB is 5.3 Å (16); S. typhimurium NrdF and mouse R2兾NrdB have a bridging water molecule, but the E. coli R2兾NrdB does not. The B. anthracis (Fig. 3e) and the M. tuberculosis (29) radicals retain their hyperfine couplings at room temperature, indicating a rigid active-site structure even at ambient temperature. Not only are the biophysical properties of B. anthracis NrdF different from other class Ib operons, but so is the phylogenetic relationship of the NrdF protein within the Gram-positive bacterial group and especially at the low G⫹C content subgroup (Fig. 6). The B. anthracis NrdF protein is clustered together with NrdFs from B. subtilis and from the pathogens Bacillus cereus, Bacillus thuringiensis, Staphylococcus aureus, and Staphylococcus epidermidis and from the phylogenetically primitive Deinococcus radiodurans. Surprisingly, other members of the Gram-positive low G⫹C content group are clustered in two other very different groups. In one of the groups, we find the atypical RNRs from the Mycoplasma and some Streptococcus species that lack some otherwise completely conserved iron ligands (8). The third group includes the majority of the low G⫹C content Gram-positives. The high G⫹C content Grampositive bacteria are clustered together with the Gram-negative bacteria. The previously characterized NrdF proteins (17–19) are clustered in this phylogenetic group. Perhaps the tight clustering of several NrdFs from pathogens with B. anthracis NrdF reflects a different tyrosyl radical geometry of the aerobically essential RNRs also in these species. Whether these pathogens also have enhanced susceptibility to some radical scavengers remains to be tested. HU is a specific inhibitor of DNA synthesis by scavenging the tyrosyl radical of all known class I RNRs. Although HU is used therapeutically with potent antiparasitic, antiviral, and antiproliferative properties (33–35), it has not yet been seriously considered as an antibacterial agent. Surprisingly, B. anthracis cells show extreme resistance to HU (LD50 of 59 mM) compared with several other pathogens like Pseudomonas aeruginosa (LD50 of 2 mM) (36), E. coli (LD50 of 3 mM) (this work), and M. tuberculosis (LD50 of 3–4 mM) (37). The extreme resistance of B. anthracis to HU may be due
