J Biol Inorg Chem DOI 10.1007/s00775-014-1121-x

ORIGINAL PAPER

The class Ib ribonucleotide reductase from Mycobacterium tuberculosis has two active R2F subunits ˚ smund K. Røhr • Marta Hammerstad • A Niels H. Andersen • Astrid Gra¨slund • Martin Ho¨gbom • K. Kristoffer Andersson

Received: 20 December 2013 / Accepted: 13 February 2014 Ó SBIC 2014

Abstract Ribonucleotide reductases (RNRs) catalyze the reduction of ribonucleotides to their corresponding deoxyribonucleotides, playing a crucial role in DNA repair and replication in all living organisms. Class Ib RNRs require either a diiron–tyrosyl radical (Y) or a dimanganese–Y cofactor in their R2F subunit to initiate ribonucleotide reduction in the R1 subunit. Mycobacterium tuberculosis, the causative agent of tuberculosis, contains two genes, nrdF1 and nrdF2, encoding the small subunits R2F-1 and R2F-2, respectively, where the latter has been thought to serve as the only active small subunit in the M. tuberculosis class Ib RNR. Here, we present evidence for the presence of an active FeIII 2 –Y cofactor in the M. tuberculosis RNR R2F-1 small subunit, supported and characterized by UV–vis, X-band electron paramagnetic resonance, and resonance Raman spectroscopy, showing features similar to those for the M. tuberculosis R2F-2– FeIII 2 –Y cofactor. We also report enzymatic activity of FeIII 2 –R2F-1 when assayed with R1, and suggest that the active M. tuberculosis class Ib RNR can use two different small subunits, R2F-1 and R2F-2, with similar activity.

˚ . K. Røhr  K. K. Andersson (&) M. Hammerstad  A Department of Biosciences, Section of Biochemistry and Molecular Biology, University of Oslo, PO Box 1066, Blindern, 0316 Oslo, Norway e-mail: [email protected] M. Hammerstad  A. Gra¨slund  M. Ho¨gbom Department of Biochemistry and Biophysics, Stockholm University, Svante Arrhenius Va¨g 16C, 106 91 Stockholm, Sweden N. H. Andersen Department of Chemistry, University of Oslo, PO Box 1033, Blindern, 0315 Oslo, Norway

Keywords Ribonucleotide reductase  R2  Tyrosyl radical  Mycobacterium tuberculosis  Iron Abbreviations EPR Electron paramagnetic resonance EU Enzyme unit HEPES N-(2-Hydroxyethyl)piperazine-N0 ethanesulfonic acid RNR Ribonucleotide reductase

Introduction Ribonucleotide reductases (RNRs) catalyze the conversion of ribonucleotides to their corresponding deoxyribonucleotides in all living organisms, providing the essential building blocks required for DNA replication and repair [1, 2]. Three main classes of RNRs have been characterized on the basis of their structure, allosteric regulation, and metallocofactors required for nucleotide reduction: classes I, II, and III. All three classes share a common radicalbased catalytic mechanism, where nucleotide reduction is initiated through the reversible one-electron oxidation of a conserved cysteine residue to a thiyl radical in the catalytic subunit. The different classes use different cofactors to generate the thiyl radical [3]. The oxygen-dependent class I RNRs, which share the same structural fold, consist of two subunits designated a and b, and contain a dimetal–oxygen cluster that in most cases is coupled to a nearby tyrosyl radical (Y) [2]. Furthermore, class I RNRs are divided into three subclasses, Ia, Ib, and Ic, on the basis of the identity of the metal cluster. In class Ia RNRs, found in eukaryotes, some viruses, and some prokaryotes such as Escherichia

123

J Biol Inorg Chem

coli, there is a l-oxo-bridged diferric tyrosyl radical (FeIII 2 – Y) cofactor. In class Ib RNRs, found in a wide variety of prokaryotes such as E. coli, Mycobacterium tuberculosis, Corynebacterium ammoniagenes, Bacillus subtilis, and Bacillus cereus [4], there may be either an FeIII 2 –Y or a MnIII 2 –Y cofactor, whereas the class Ic RNR, so far only well characterized in the obligate intracellular pathogen Chlamydia trochomatis, utilizes a stable, heterodinuclear MnIVFeIII cofactor to initiate catalysis [5, 6]. The class II RNRs use a cobalt-containing deoxyadenosylcobalamin (vitamin B12) as a radical-generating cofactor [7], whereas class III RNRs have a glycyl radical generated by an iron– sulfur cluster together with S-adenosylmethionine [3, 8]. M. tuberculosis, the infectious agent causing tuberculosis, contains a class Ib RNR system, encoded by the nrdE and nrdF genes, coding for the large, catalytic subunit (R1E/NrdE) and the small, cofactor-housing subunit (R2F/ NrdF), respectively, in addition to the putative class II RNR, encoded by nrdZ [9, 10]. The simultaneous presence of more than one class of RNRs in some organisms may be due to adaptation to different O2 levels in the environment [11]. Further complexity is also provided in the case of M. tuberculosis class Ib RNR by the presence of two genes, nrdF1 and nrdF2, encoding the class Ib small subunits (R2F-1 and R2F-2, respectively) [9, 10]. Previous studies have shown that R2F-2 containing the FeIII 2 –Y cofactor is enzymatically active when assayed with the M. tuberculosis large subunit R1, hence being able to form a functional class Ib RNR [9, 10, 12]. In contrast, the alternate class Ib RNR small subunit R2F-1 has been suggested to be unable to substitute for R2F-2 in the active class Ib RNR system. As previous attempts at characterizing R2F-1 at the protein level, performed by Yang et al. [9], resulted in insoluble cytoplasmic inclusion bodies, biochemical investigation of the M. tuberculosis class Ib RNR has only been performed using purified R2F-2 [9]. The absence of soluble R2F-1 protein in the latter study could be a consequence of the high temperatures used for gene expression. On the basis of these results, as well as inhibition studies performed with C-terminal R2F peptides, R2F-2 was assumed to be the only active R2F subunit in M. tuberculosis [9]. The two nrdF genes coding for the R2F subunits, nrdF1 and nrdF2, have 71 % identity and 83 % sequence similarity at the amino acid level, and are both homologous with other class Ib RNRs [9]. Additionally, both M. tuberculosis R2F subunits have been shown to contain the conserved tyrosine residue responsible for active tyrosyl radical generation, together with the conserved metal-binding residues where the dimetal clusters are coordinated by four carboxylates and two histidines [13]. The residues responsible for forming the hydrophobic pocket into which the l-oxo bridge of the FeIII 2 –Y cofactor can project are also conserved in the two M. tuberculosis R2Fs [9, 12].

123

It has been shown that the nrdF1 gene is expressed at a level comparable to that of nrdF2, suggesting that it may serve a cellular role; however, expression of nrdF1 and nrdF2 is not controlled in a similar fashion [10]. Although several genetic studies have been performed on the class Ib RNR from M. tuberculosis [9, 10, 14, 15], they have provided sufficient biochemical and biophysical evidence for the biological role and characterization of R2F-1. Here, we present spectroscopic evidence for the presence of an active FeIII 2 –Y cofactor in the M. tuberculosis class Ib RNR R2F-1 small subunit, supported by UV–vis and X-band electron paramagnetic resonance (EPR) spectroscopy, showing characteristics similar to those previously observed for M. tuberculosis R2F-2 [12]. Further, we present the first characteristics of the tyrosyl radical in both M. tuberculosis class Ib RNR small subunits, R2F-1 and R2F-2, performed by resonance Raman spectroscopy. Most importantly, we report enzymatic activity of R2F-1 containing the FeIII 2 –Y cofactor when assayed with R1, showing specific activity similar to that of the R2F-2 subunit, suggesting that R2F-1 is able to substitute for R2F2 in the active M. tuberculosis class Ib RNR.

Materials and methods Cloning, expression, and purification N-terminally tagged His6-R1, His6-NrdI, His6-R2F-1, and His6-R2F-2 from M. tuberculosis M. tuberculosis H37Rv R1 (Rv3051c), NrdI (Rv3052c), R2F-1 (Rv1981c), and R2F-2 (Rv3048c) [16] were cloned by PCR, and the constructs were inserted into the pET-46 Ek/LIC vector (Novagen), introducing an N-terminal His6tag. All constructs were sequenced to confirm the correct inserts prior to transformation into competent E. coli BL21 (DE3) Gold cells (Stratagene). Cells containing pET-46 Ek/LIC-R1, pET-46 Ek/LIC-NrdI, pET-46 Ek/LIC-R2F-1, or pET-46 Ek/LIC-R2F-2 were grown in terrific broth containing 100 lg mL-1 ampicillin. Protein expression was induced by adding isopropyl b-D-1-thiogalactopyranoside to a final concentration of 0.5 mM, and the cultures were incubated for 12–16 h at 291 K with vigorous shaking before harvesting. Cells containing pET-46 Ek/LICR2F-2 were, in addition, grown in medium supplemented with 100 lM 1,10-phenanthroline 20 min prior to induction. The proteins were lysed in a French press and purified by nickel nitrilotriacetic acid agarose (Qiagen) and sizeexclusion chromatography (HiLoad 16/60 Superdex 200, GE Healthcare). His6-R1 and His6-NrdI were stored in 50 mM tris(hydroxymethyl)aminomethane–HCl, pH 7.5 and 1 mM dithiothreitol, whereas His6-R2F-1 and His6-

J Biol Inorg Chem

R2F-2 were stored in 50 mM N-(2-hydroxyethyl)piperazine-N0 -ethanesulfonic acid (HEPES), pH 7.5 and 200 mM Na2SO4. M. tuberculosis R2F-1 To increase protein solubility, M. tuberculosis R2F-1 was also cloned, expressed, and purified without an N-terminal His6-tag. Gene expression and protein purification were performed as described previously [17, 18]. In brief, the nrdF1 gene cloned into a pET-22b plasmid was ordered from GenScript. The vector was then transformed into competent E. coli BL21 (DE3) Gold cells (Stratagene). Cells containing pET-22b-R2F-1 were grown in terrific broth containing 100 lg mL-1 ampicillin. Protein expression was induced by adding isopropyl b-D-1-thiogalactopyranoside to a final concentration of 0.8 mM, and the culture was incubated for 12–16 h at 293 K with vigorous shaking before harvesting. The frozen cell paste was lysed using an X-pressÒ [19]. The R2F-1 protein was purified by ion-exchange (HiTrap HP Q, GE Healthcare) and size-exclusion (Superdex 200, GE Healthcare) chromatography, and stored in 50 mM HEPES, pH 7.5 and 200 mM Na2SO4. The two R2F subunits show differences in the level of expression and stability during protein purification, where R2F-1 has been shown to be less soluble and give a lower protein yield than R2F-2. However, the N-terminally tagged His6-R2F-1 showed an even lower solubility compared with the untagged R2F-1 protein, the latter being used for characterization. Reconstitution of the tyrosyl radical in M. tuberculosis R2F-1 and R2F-2 After purification, R2F-1 and R2F-2 proteins (when expressed without 1,10-phenanthroline) were treated with a metal chelator solution containing 8-hydroxyquinoline, as described previously by Atkin et al. [20] with some modifications, and were passed through a 5 mL HiTrap desalting column (GE Healthcare) to remove iron and to quench the tyrosyl radical before reconstitution with FeII. Protein reactivation of apo-R2F-1 or apo-R2F-2 was performed with addition of excess FeII and O2, using a solution of (NH4)2Fe(SO4)26H2O, pH * 2.3, giving a ratio of seven FeII atoms to one R2F. UV–vis spectroscopy The active M. tuberculosis R2F-1 and R2F-2 proteins were analyzed by UV–vis spectroscopy using either a JASCO V-560 UV–vis spectrophotometer or an Agilent 8453 diode-array UV–vis spectrophotometer.

EPR and Raman sample preparation Samples of M. tuberculosis R2F-1 or R2F-2 (150– 200 lL) in 50 mM HEPES, pH 7.5, 200 mM Na2SO4/ 100 mM NaCl and 10 % glycerol, were prepared for EPR or Raman experiments. In addition, samples of active M. tuberculosis R2F-1 or R2F-2 in deuterated buffer (50 mM HEPES and 200 mM Na2SO4/100 mM NaCl, pD 7.9) were made by adding apo-R2F-1 or apo-R2F-2 onto a 5 mL HiTrap desalting column (GE Healthcare) equilibrated with deuterated buffer (D2O, 99.9 % D, Cambridge Isotope Laboratories) before reconstitution of the diiron–oxygen cluster as explained earlier. All samples were transferred to quartz EPR tubes and frozen in liquid nitrogen. X-band EPR experiments X-band EPR spectra of M. tuberculosis R2F-1 and R2F-2 between 4 and 50 K were recorded using a Bruker Elexsys 560 EPR spectrometer equipped with an Oxford Instruments ESR900 helium flow cryostat, and a Bruker ER41116DM dual-mode cavity or a Super X kv319 cavity. All samples contained 10 % (v/v) glycerol for vitrification during the low-temperature recordings. EPR spectra were measured at different microwave powers, and quantified by double integration of the spectra recorded under nonsaturating conditions, compared with standards of 50 mM HEPES, pH 7.5, and 500 lM CuIISO4 in HClO4 (0.4 Y per R2F-2, consistent with quantifications reported previously [12] ). First-derivative spectra were recorded at different microwave powers (P) and at various temperatures to determine the microwave power at half saturation (P1/2) for each temperature examined. The data were fitted as described previously [21–24]. Simulation of EPR spectra EPR spectra of M. tuberculosis R2F-1 and R2F-2 were simulated using the EasySpin package [25]. The experimental spectra are nearly identical; thus, the previously published [12] R2F-2 g tensor, which was determined using high-field EPR spectroscopy, and proton hyperfine splitting tensors were used as initial simulation parameters. Although the g tensor was fixed during the simulations, both R2F-1 and R2F-2 were simulated assuming three A tensors (AHb, AH3,5, and AH2,6). Both spectra were simulated using the esfit module (Levenberg/Marquardt mode), with first-order perturbation theory for resonance energies, and a gstrain model to fit the line widths.

123

J Biol Inorg Chem

Resonance Raman spectroscopy

Results and discussion

A HORIBA Jobin Yvon T64000 instrument operating in single-grating mode was used for resonance Raman spectroscopy. A Kaiser Optical Systems holographic SuperNotch filter (410 nm) ensured effective Rayleigh light retention, and the spectral width of 2.56 cm-1 was obtained by setting the entrance slit width to 100 lm and use of a grating with 2,400 grooves per millimeter. Each Raman spectrum was a result of averaging 40 scans of 90 s. Samples were kept in EPR tubes cooled with liquid N2 in a cold-finger EPR cryostat made of quartz. Fluorescence was subtracted by fitting the spectra with moderate-degree polynomial expressions, and Raman frequencies were calibrated against well-known spectra of 4-acetamidophenol (tylenol) [26]. Only the samples were moved during the experiment, and the spectrograph was kept fixed all the time. The final accuracy achieved was about 0.5 cm-1. Laser light was provided by a three-stage system consisting of a Spectra Physics Millennia Pro 12sJ Nd:YVO4 laser serving as a pump laser (6 W at 532.1 nm) for a MatisseTR Ti:sapphire ring laser emitting approximately 1 W at 820 nm that was finally converted to 410 nm through a WaveTrain frequency doubler yielding 20 mW, which was attenuated to 2 mW at the sample via the optical parts.

UV–vis spectroscopy of iron-reconstituted M. tuberculosis R2F-1 and R2F-2

Activity measurements of M. tuberculosis R2F-1 and R2F-2 with R1 Activity assay conditions were chosen on the basis of work previously described by Yang et al. [9], with some modifications. Activities were determined from the rate of reduction of [3H]cytidine diphosphate. With optimized assay conditions, M. tuberculosis class Ib RNR activity for the two different small subunits, R2F-1 and R2F-2, was determined in 100 lL reaction mixtures containing 50 mM tris(hydroxymethyl)aminomethane–HCl, pH 7.5, 20 mM MgCl2, 1.8 mM adenosine triphosphate, 8 mM dithiothreitol, and 0.2 mM [3H]cytidine diphosphate (98,200 cpm min-1 nmol-1). R2F-1 or R2F-2 (2 lg ) was titrated with R1 (0–60 lg), and the reaction mixtures were incubated at 310 K for 10 min, the reactions were terminated at 369 K for 5 min, and the reaction mixtures were cooled on ice and centrifuged at 14,000g. The samples were applied to columns packed with Affi-Gel boronate affinity gel (Bio-Rad Laboratories), regenerated with 0.1 M sodium boronate pH 8.9, and equilibrated with 15 mM MgCl2 and 50 mM (NH4)2CO3/NH4HCO3, pH 8.9. The 20 -deoxycytidine diphosphate formed was analyzed as described previously [27], where one enzyme unit (EU) is defined as 1 nmol of 20 -deoxycytidine diphosphate produced per minute, and specific activity is expressed as EU per milligram of R2F-1 or R2F-2 protein.

123

Both M. tuberculosis apo-R2F-1 and M. tuberculosis apoR2F-2 can be reconstituted with excess ferrous iron and O2 to the FeIII 2 –Y state, but only the latter was described previously [9, 12]. However, the iron-loaded R2F-1–Y was considerably less stable than the corresponding R2F2–Y, which is likely the reason for a minor degree of protein precipitation at room temperature, leading to increased light scattering and a lower Y and diiron oxygen cluster yield in R2F-1 as compared with R2F-2 (Fig. 1). The UV–vis spectra of the 10 lM iron-reconstituted R2F-1 and R2F-2 proteins both display peaks at 408 nm, characteristic of a tyrosyl radical (Fig. 1). The maximum yield of Y was obtained with 7 equiv of FeII per R2F dimer. The radical content in the UV–vis spectroscopy protein samples was estimated to be 0.15 Y per R2F-1 and 0.25 Y per R2F-2 on the basis of the subtracted radical peak spectra, using e410 = 3.9 mM-1 cm-1 [28]. Low-temperature EPR studies of M. tuberculosis R2F-1 and R2F-2 The tyrosyl radicals in the iron-reconstituted samples of M. tuberculosis R2F-1 and R2F-2 were analyzed by lowtemperature EPR spectroscopy. X-band (9.4-GHz) absorption derivative spectra were recorded over a temperature range from 3.8 to 20 K for both proteins. The experimental and simulated spectra of R2F-1 and R2F-2 are shown in Fig. 2 (X-band, 20 K). As usual, hyperfine splittings from several protons dominate the spectral

Fig. 1 UV–vis spectra of 10 lM R2F-1 (brown line) and 10 lM R2F-2 (blue line) reconstituted with ferrous iron at room temperature (1 cm cuvette length). The inset shows the Y peaks corrected by an ad hoc parabolic function for R2F-1 and R2F-2

J Biol Inorg Chem

tensor components as R2F-2. The X-band EPR microwave power saturation properties of R2F-1 at 3.8 K [data not shown, b values equal to 1, P1/2(3.8 K) = 0.07 mW] indicate relaxation properties of the radical system consistent with those of previously described R2 proteins [2, 32, 33]. The X-band EPR spectra of M. tuberculosis R2F-1 and R2F-2 were simulated on the basis of previously published hyperfine coupling parameters for M. tuberculosis R2F-2 [12]. The M. tuberculosis R2F-1 and R2F-2 X-band spectra from this work are nearly identical and display well resolved, nearly identical anisotropic hyperfine coupling parameters (Table 1), strongly resembling the previously published R2F-2 X-band EPR spectra. On the basis of calculations performed by Himo et al. [34], and the corresponding diagram showing the relationship between the hyperfine coupling parameter tensors of the b protons and the rotation angle, we have estimated the dihedral angle h for the tyrosyl radical in both R2F-1 and R2F-2 to be around 75°. This value is similar to the h angle previously reported for M. tuberculosis R2F-2 [12] and Salmonella typhimurium [35], in agreement with similar EPR spectral features [35] (Table 1). Resonance Raman spectroscopy

Fig. 2 X-band electron paramagnetic resonance (EPR) spectra of III Mycobacterium tuberculosis R2F-1–FeIII 2 –Y and R2F-2–Fe2 –Y. a Spectra of both samples (75 and 100 lM for R2F-1 and R2F-2, respectively, where the intensity of the latter has been multiplied by a factor of 0.75 to compensate for the protein concentration difference) (solid lines) were recorded at 20 K under nonsaturated conditions [32 lW microwave power for R2F-1 (brown line) and 40 lW microwave power for R2F-2 (blue line)], with a modulation amplitude of 0.2 mT. R2F-1 was recorded with 9.38 GHz microwave frequency, whereas R2F-2 was recorded with 9.40 GHz microwave frequency. Simulations for both spectra are shown as dashed lines, with corresponding colors as described. The spectrum simulations show the EPR line when contributions from the anisotropic hyperfine terms due to the two included H3, H5 protons, the Hb1 proton, and the two H2, H6 protons. b An overlay of the two corresponding R2F-1 and R2F-2 EPR spectra, illustrating the nearly identical features

envelope. Minor differences in the radical resonance envelope were observed when comparing R2F-1–FeIII 2 –Y with R2F-2–FeIII 2 –Y. The anisotropic g-tensor components are not resolved at X-band frequency; however, they have previously been determined for R2F-2 using high-field EPR spectroscopy at 285 GHz (gx = 2.0092, gy = 2.0046, and gz = 2.0022) [12, 29]. The high gx value observed for R2F-2 indicates no hydrogen bonding to the tyrosyl-oxygen radical [30, 31], in agreement with Raman spectroscopy studies performed on both R2F-2 and R2F-1 (see below), indicating that the proteins are similar. Therefore, we assume that the tyrosyl radical in R2F-1 has the same g-

Raman spectroscopy makes possible characterization of structural details via detection of changes in spectral patterns, band frequencies, band widths, and intensities. In this case, the vibrational characteristics of tyrosyl radicals are sensitive to the presence of exchangeable acidic hydrogens located close to the phenoxyl oxygen. Laser excitation near the absorption maximum of tyrosyl radicals, which typically falls around 410 nm, causes a significant enhancement of the Raman signals associated with C–O stretching [36]. This concerns particularly the v~7a mode (Wilson notation for benzene) [37], making possible detection of changes in out-of-plane bond angles, as well as changes in the local charge and dielectric environment affecting the radical state. The vibrational mode of major interest, v~7a , is observed near 1,496 cm-1 for both R2F-1 and R2F-2 (Fig. 3, Table 2). Typically, v~7a falls within the 1,495–1,500 cm-1 range when no phenolate oxygen hydrogen bond is present. However, in the presence of a hydrogen bond, also shown by electron–nuclear double resonance [38], as in the case of mouse R2, the v~7a mode shifts to higher levels (1,515 cm-1). The latter study also reported a 5 cm-1 downshift of the mouse R2 protein when reconstituted in D2O (Table 2), as discussed in more detail in [2, 39]. In the case of R2F-1, no such deuterium shift is observed, whereas an almost negligible downshift of 1 cm-1 appears for R2F-2. This indicates that either there is no exchangeable hydrogen adjacent to the radical or that a possible proton cannot exchange under this experimental

123

J Biol Inorg Chem Table 1 The g tensors, hyperfine coupling parameter tensors (A, mT), line widths (LW; mT), and rotational h angle (°) for FeIII 2 –Y in Mycobacterium tuberculosis class Ib ribonucleotide reductases (RNRs)

EPR parameters

Class Ib RNRs M. tuberculosis R2F-1

M. tuberculosis R2F-2

M. tuberculosis R2F-2 [12]

S. typhimurium R2F [35]

gx

2.0092

2.0092

2.0092

2.0089

gy

2.0046

2.0046

2.0046

2.0043

gz

2.0022

2.0022

2.0022

2.0021

Ax

1.19

1.15

0.89

1.05

Ay

1.02

0.92

0.73

0.74

Az Hb2

1.04

0.93

0.90

0.95

Ax

–

–

–

\0.30

Ay

–

–

–

\0.30

Az

–

–

–

\0.30

Ax

-0.87

-0.86

-1.17

-1.15

Ay

0.009

-0.26

-0.19

-0.25

Az

-0.74

-0.70

-0.71

-0.71

Ax

0.23

0.007

0.28

–

Ay

0.66

0.61

0.20

–

Az

Hb1

H3,5

H2,6

EPR electron paramagnetic resonance a

g-strain values

0.11

0.10

0.11

–

LWx

0.0020a

0.0031a

0.58

0.58

LWy

0.0045a

0.0036a

0.54

0.53

LWz h

a

0.0015a 75

0.29 75

0.37 75

0.0015 75

The weaker band observed at 1,553 cm-1 is due to dioxygen that tends to accumulate in the cryostat, and contamination of solid CO2 is evident from the modes at 1,384 cm-1 and 1,284 cm-1 (not shown). A broad band associated with CH2 deformation in glycerol appears around 1,466 cm-1 [40], and the band at 1,405 cm-1 most likely is caused by the presence of tyrosinate [41]. Enzyme activity

Fig. 3 Resonance Raman spectra of M. tuberculosis R2F-1–FeIII 2 –Y and R2F-2–FeIII 2 –Y. The resonance signature of the tyrosyl radical v~7a vibration for R2F-1 (brown lines) and R2F-2 (blue lines), dissolved in H2O and D2O buffers. The conditions were as follows: excitation wavelength 410 nm, laser power at the sample 2 mW, temperature 77 K. The asterisks indicate artifact signatures of solid CO2 and liquid O2, respectively. A downshift of 1 cm-1 is observed for R2F-2 recorded in D2O

condition. However, the most likely explanation is the first, as also observed for the E. coli (1,497–1,501 cm-1) [36] and B. cereus (1,500 cm-1) [21] class Ia R2s (Table 2).

123

Under conditions where R2F-1 or R2F-2 is the limiting species (2 lg), with a saturating amount of R1 (R1 concentration approximately 9 times higher than the R2F concentration ), the specific activity of the R2F-1 subunit was 180 (s ± 48) EU mg-1 R2F-1, whereas the specific activity of the R2F-2 subunit was 170 (s ± 18) EU mg-1 R2F-2. These results show high and similar specific activity for both R2F-1 and R2F-2 under standard conditions, consistent with the previously reported activity for the R2F-2 subunit [9]. Importantly, these results are the first to demonstrate catalytic activity of the second putative M. tuberculosis class Ib R2F-1 subunit when assayed with the large subunit, R1. The relative large standard deviation calculated for the specific activity of R2F-1 can likely be

J Biol Inorg Chem Table 2 Resonance Raman parameters for the phenoxyl v~7a bands of different RNRs

Phenoxyl v~7a

Class Ia or class Ib RNR protein Mouse R2 [54]

E. coli class Ia R2 [36]

B. cereus R2F [55]

M. tuberculosis R2F-1

M. tuberculosis R2F-2

v (cm-1)

1,515

1,498

1,500

1,496

1,496

DD (cm-1)

-5

0

0

0

-1

explained by R2F-1 instability for this protein, as also observed during protein purification and active R2F-1– FeIII 2 –Y generation. The degree of protein precipitation at the low protein concentrations used for the activity measurements is assumed to be low. However, if a minor degree of R2F-1 protein precipitation is present, this could cause the specific activity to appear lower than that under optimal experimental conditions. Comparison of R2F-1 and R2F-2 M. tuberculosis R2F-1 and R2F-2 are homologous, with 71 % sequence identity, 83 % sequence similarity, and similar amino acid sequence length (324 residues in R2F-2 and 322 residues in R2F-1). An alignment of the R2F-1 and R2F-2 sequences indicates that the metal-binding residues are identical in the two R2F proteins, and are thus most likely structurally very similar to R2F-2, where the dimetal clusters are coordinated by four carboxylates and two histidines, as for most class Ia and Ib RNRs [13] (Fig. 4a). Furthermore, the sequence alignment suggests a high level of structural similarity in the vicinity surrounding the dimetal sites and the coordinating residues, as also revealed by the homology model of R2F-1, compared with the structure of R2F-2 (data not shown) [42]. Owing to differences in the amino acid composition in the C-terminal tail of R2F-1 and R2F-2, it has previously been suggested that R2F-2 is the only complementing subunit that can function to form an active holoenzyme complex with R1 [9, 42]. The final model of the R2F-2 crystal structure solved by Uppsten et al. [42] lacks the terminal 33 residues owing to disorder. A disordered C-terminus appears to be a common feature among the R2 proteins. This is likely a functional feature as it has been shown that the C-terminal part of the R2 proteins interacts with the R1 protein in the active holoenzyme [43–45]. Sequence alignment demonstrates a 70 % sequence identity among these 33 residues between R2F-1 and R2F-2. Inhibition studies using C-terminal peptides analogous to the seven C-terminal residues of R2F-1 and R2F-2 have previously demonstrated that peptides corresponding to the C-terminal tail of R2F-1 inhibit RNR activity (R2F-2 assayed with R1) to a fivefold lower extent than peptides corresponding to the C-terminal tail of R2F-2 [9]. Hence, it has been suggested that the R2F-1 subunit would have to be present in concentrations

Fig. 4 a The metal ion site in M. tuberculosis R2F-2. The R2F-2 reduced iron coordination site (monomer A) (Protein Data Bank ID 1UZR) [42], with the iron ions displayed as orange spheres, and the coordinating amino acids and the tyrosine responsible for Y generation represented as sticks and colored by atom type. The metal-coordinating residues and tyrosine are also conserved in M. tuberculosis R2F-1, showing high structural similarity to the M. tuberculosis R2F-2 structure. b Residues on the surface of the class Ib M. tuberculosis R2F-2 (Protein Data Bank ID 1UZR) [42], B. cereus R2F (Protein Data Bank ID 4BMP) [50], and E. coli R2F (Protein Data Bank ID 3N39) [51], representing the conserved core residues shown to be involved in the hydrogen-bonding network in the NrdI–R2F interface in the class Ib ribonucleotide reductase. The conserved residues are also present in M. tuberculosis R2F-1, showing high structural similarity to the corresponding M. tuberculosis, B. cereus, and E. coli R2F structures. The figure was prepared using PyMOL [53]

higher than the R2F-2 subunit in order to form an active a2b2 holoenzyme complex. The seven C-terminal residues in the two R2Fs differ by only two amino acids, where Glu-

123

J Biol Inorg Chem

318 and Asp-320 in R2F-2 have been replaced with Thr316 and Thr-318, respectively, in R2F-1. The loss of the two carboxylates in R2F-1 might result in changes in the R2F–R1 affinity, as previously suggested [9]. However, such peptide inhibition studies had never previously been performed on an active R2F-1–R1 complex, and had only been performed on the active R2F-2–R1 complex [9]. As demonstrated in our study, similar specific activity reported for the R2F-1–R1 and R2F-2–R1 complexes suggests that the affinity of R2F-1 for R1 must be substantial to form an active holoenzyme complex. This also suggests that peptide inhibition of the active R2F-2–R1 complex with C-terminal peptides corresponding to the different R2Fs does not provide adequate information about R2F–R1 binding and activity with respect to the different R2Fs, providing limited information compared with activity measurements performed using purified full-length proteins, as performed in this study. Even though R2F-2–FeIII 2 – Y has been shown to be stabler than R2F-1–FeIII –Y at 2 higher concentrations, similar catalytic activity for both R2Fs suggests that the M. tuberculosis class Ib enzyme can use both R2F-2 and R2F-1 under similar in vitro conditions. Previous studies suggesting that R2F-2 is the only in vivo active small subunit in M. tuberculosis were not based on protein characterization. However, genetic studies performed by Mowa et al. [15] have shown that loss of the nrdF1 gene in M. tuberculosis does not make the mutants more sensitive to drugs or UV irradiation, nor does it affect the virulence of M. tuberculosis as compared with the wild-type strains [15]. The same study also showed that the loss of nrdF1 affected the expression of nrdF2 to a small extent; however, the effect of an nrdF2 deletion mutant on nrdF1 expression was not investigated. Also, different experimental conditions were not examined, which could provide additional information. Studies performed by Dawes et al. [10] have shown that nrdF2 is essential for growth of M. tuberculosis (on the basis of knockout of the gene), and that it is expressed at a level comparable to that of nrdF1 under low O2-level tension [10]. On the basis of the latter results and findings from this study, we suggest that both R2F subunits can be used in the class Ib RNR in M. tuberculosis. However, a more extensive characterization comparing the roles of R2F-1 and R2F-2 as functional M. tuberculosis class Ib RNR Y generating subunits is still required. A 3D structure of R2F-1 would also provide useful information regarding the function of this protein. The MnIII 2 –Y cofactor in M. tuberculosis class Ib RNR An increasing number of studies have reported that class Ib RNRs containing the MnIII 2 –Y cofactor has a higher specific activity compared with the iron form of the enzyme [46, 47], and that the MnIII 2 –Y cofactor can be formed both

123

in vitro and in vivo [21, 48]. However, specific activity for the M. tuberculosis class Ib RNR containing the active MnIII 2 –Y cofactor has never been confirmed. Our attempts to reconstitute the active MnIII 2 –Y cofactor in M. tuberculosis R2F-1 and R2F-2 in the presence of M. tuberculosis NrdI using previously described procedures [47, 49] were unsuccessful (data not shown). Nonetheless, the presence of NrdI in the M. tuberculosis genome suggests that this bacterium could also be able to generate a MnIII 2 – Y cofactor in both small R2F subunits. A 3D structural alignment of the M. tuberculosis R2F-2 crystal structure with the structures of the class Ib R2F subunits of B. cereus and E. coli from the NrdI–Fe2–R2F and NrdI–MnII2 –NrdF crystal complexes [50, 51] reveals a high level of conservation of the amino acids lining the R2F interface in the NrdI–R2F binding site between M. tuberculosis, B. cereus, and E. coli, including the charged and polar residues suggested to create a hydrophilic environment used for channeling of O2- needed for active MnIII 2 –Y cofactor generation [52]. A comparison of the M. tuberculosis R2F2 and B. cereus and E. coli NrdI–R2F structures highlighting the conserved residues lining the NrdI–R2F core binding interface in the R2s is shown in Fig. 4b. The conserved residues lining the R2F binding site are identical in R2F-1 and in R2F-2, as seen from sequence alignment, indicating a close structural similarity. This core interface is likely essential for recognition between R2s and NrdIs in class Ib RNRs. This proposes a similar NrdI binding to the R2F-1 and R2F-2 surface in M. tuberculosis class Ib RNR, required for MnIII 2 –Y cofactor generation, as seen in previously solved NrdI–R2F crystal complex structures [50, 51]. The simultaneous presence of two R2F subunits in the class Ib RNR in M. tuberculosis could also play a role in discrimination of metal ion incorporation during distinct growth conditions, varying between the two R2F subunits and distinguishing between generation of a MnIII 2 –Y cofactor and an FeIII –Y cofactor. Incorporation of different 2 metal ions could be a consequence of variable O2 levels or metal status, and could possibly influence the viability of cells in different habitats.

Conclusion In this study, we expressed and purified the previously uncharacterized and previously assumed inactive M. tuberculosis class Ib RNR alternate R2F-1 subunit, and demonstrated catalytic activity for R2F-1 when assayed with R1. The results were compared with those for the well-characterized R2F-2 subunit, showing similar specific activity. Furthermore, we used UV–vis, X-band EPR, and resonance Raman spectroscopy to investigate the ironreconstituted tyrosyl radicals from both R2F-1 and R2F-2,

J Biol Inorg Chem

revealing nearly identical features. These findings suggest that M. tuberculosis can alternate between two radicalgenerating subunits in the class Ib RNR under the same in vitro conditions, differing from other characterized organisms expressing this class of RNR. However, further investigations are required for a deeper understanding of the biological relevance of the presence of two catalytically active subunits. The conserved metal ion coordination environment, investigated spectroscopically, supports the generation and stabilization of an FeIII 2 –Y cofactor in R2F-1 analogous to the previously characterized Y in R2F-2. A sequence alignment of the R2F-1 and R2F-2 surface residues with those of the E. coli [51] and B. cereus structures of R2F in complex with NrdI also reveals the presence of common conserved residues suggested to serve in the recognition between R2s and NrdIs in class Ib RNRs. This opens up the possibility of active MnIII 2 –Y cofactor generation in both M. tuberculosis class Ib small subunits. Acknowledgments This work was financially supported by the Norwegian Research Council through projects 214239/F20 and 218412/F50 (K.K.A.) and CMST COST Action CM1003 (K.K.A. and A.G.), the Swedish Research Council [2010-5061 (M.H.) and 2011-4850 (A.G.)], the Swedish Foundation for Strategic Research, and the Knut and Alice Wallenberg Foundation (M.H.) Travel grants were provided by the MLSUiO program for molecular life science research at the University of Oslo, and the National Graduate School in Structural Biology (BioStruct) (UiT/NorStruct). We thank HansPetter Hersleth (University of Oslo) for useful discussions.

References 1. Nordlund N, Reichard P (2006) Annu Rev Biochem 75:681–706 2. Tomter AB, Zoppellaro G, Andersen NH, Hersleth HP, Hammerstad M, Rohr AK, Sandvik GK, Strand KR, Nilsson GE, Bell CB, Barra AL, Blasco E, Le Pape L, Solomon EI, Andersson KK (2013) Coord Chem Rev 257:3–26 3. Andersson KK (ed) (2008) Ribonucleotide reductase. Nova, Hauppauge 4. Lundin D, Torrents E, Poole AM, Sjoberg BM (2009) BMC Genom 10:589 5. Voevodskaya N, Lendzian F, Ehrenberg A, Graslund A (2007) FEBS Lett 581:3351–3355 6. Jiang W, Yun D, Saleh L, Barr EW, Xing G, Hoffart LM, Maslak MA, Krebs C, Bollinger JM (2007) Science 316:1188–1191 7. Booker S, Stubbe J (1993) Proc Natl Acad Sci USA 90:8352–8356 8. Fontecave M, Eliasson R, Reichard P (1989) J Biol Chem 264:9164–9170 9. Yang FD, Curran SC, Li LS, Avarbock D, Graf JD, Chua MM, Lui GZ, Salem J, Rubin H (1997) J Bacteriol 179:6408–6415 10. Dawes SS, Warner DF, Tsenova L, Timm J, McKinney JD, Kaplan G, Rubin H, Mizrahi V (2003) Infect Immun 71:6124–6131 11. Poole AM, Logan DT, Sjoberg BM (2002) J Mol Evol 55:180–196 12. Liu A, Potsch S, Davydov A, Barra AL, Rubin H, Graslund A (1998) Biochemistry 37:16369–16377

13. Solomon EI, Brunold TC, Davis MI, Kemsley JN, Lee SK, Lehnert N, Neese F, Skulan AJ, Yang YS, Zhou J (2000) Chem Rev 100:235–349 14. Nurbo J, Roos AK, Muthas D, Wahlstrom E, Ericsson DJ, Lundstedt T, Unge T, Karlen A (2007) J Pept Sci 13:822–832 15. Mowa MB, Warner DF, Kaplan G, Kana BD, Mizrahi V (2009) J Bacteriol 191:985–995 16. Cole ST, Brosch R, Parkhill J, Garnier T, Churcher C, Harris D, Gordon SV, Eiglmeier K, Gas S, Barry CE, Tekaia F, Badcock K, Basham D, Brown D, Chillingworth T, Conner R, Davies R, Devlin K, Feltwell T, Gentles S, Hamlin N, Holroyd S, Hornsby T, Jagels K, Krogh A, McLean J, Moule S, Murphy L, Oliver K, Osborne J, Quail MA, Rajandream MA, Rogers J, Rutter S, Seeger K, Skelton J, Squares R, Squares S, Sulston JE, Taylor K, Whitehead S, Barrell BG (1998) Nature 396:190–198 17. Tomter AB, Bell CB, Rohr AK, Andersson KK, Solomon EI (2008) Biochemistry 47:11300–11309 18. Rohr AK, Hersleth HP, Andersson KK (2010) Angew Chem Int Ed 49:2324–2327 19. Magnusson KE, Edebo L (1976) Biotechnol Bioeng 18:865–883 20. Atkin CL, Thelande L, Reichard P, Lang G (1973) J Biol Chem 248:7464–7472 21. Tomter AB, Zoppellaro G, Bell CB, Barra AL, Andersen NH, Solomon EI, Andersson KK (2012) PLoS One 7:e33436 22. Castner TG (1959) Phys Rev 115:1506–1515 23. Portis AM (1953) Phys Rev 91:1071–1078 24. Zoppellaro G, Harbitz E, Kaur R, Ensign AA, Bren KL, Andersson KK (2008) J Am Chem Soc 130:15348–15360 25. Stoll S, Schweiger A (2006) J Magn Reson 178:42–55 26. McCreery RL (2000) Raman spectroscopy for chemical analysis. Wiley, New York 27. Thelander L, Sjo¨berg B-M, Eriksson S (1978) Methods Enzymol 51:227–237 28. Petersson L, Graslund A, Ehrenberg A, Sjoberg BM, Reichard P (1980) J Biol Chem 255:6706–6712 29. Liu AM, Barra AL, Rubin H, Lu GZ, Graslund A (2000) J Am Chem Soc 122:1974–1978 30. Andersson KK, Schmidt PP, Katterle B, Strand KR, Palmer AE, Lee SK, Solomon EI, Graslund A, Barra AL (2003) J Biol Inorg Chem 8:235–247 31. Svistunenko DA, Jones GA (2009) Phys Chem Chem Phys 11:6600–6613 32. Sahlin M, Petersson L, Graslund A, Ehrenberg A, Sjoberg BM, Thelander L (1987) Biochemistry 26:5541–5548 33. Galli C, Atta M, Andersson KK, Graslund A, Brudvig GW (1995) J Am Chem Soc 117:740–746 34. Himo F, Graslund A, Eriksson LA (1997) Biophys J 72:1556–1567 35. Allard P, Barra AL, Andersson KK, Schmidt PP, Atta M, Graslund A (1996) J Am Chem Soc 118:895–896 36. Backes G, Sahlin M, Sjoberg BM, Loehr TM, Sandersloehr J (1989) Biochemistry 28:1923–1929 37. Wilson EB (1934) Phys Rev 45:706–714 38. van Dam PJ, Willems JP, Schmidt PP, Potsch S, Barra AL, Hagen WR, Hoffman BM, Andersson KK, Graslund A (1998) J Am Chem Soc 120:5080–5085 39. Tomter AB, Zoppellaro G, Schmitzberger F, Andersen NH, Barra AL, Engman H, Nordlund P, Andersson KK (2011) PLoS One 6:e25022 40. Mendelovici E, Frost RL, Kloprogge T (2000) J Raman Spectrosc 31:1121–1126 41. Johnson CR, Ludwig M, Asher SA (1986) J Am Chem Soc 108:905–912 42. Uppsten M, Davis J, Rubin H, Uhlin U (2004) FEBS Lett 569:117–122

123

J Biol Inorg Chem 43. Climent I, Sjoberg BM, Huang CY (1991) Biochemistry 30:5164–5171 44. Fisher A, Yang FD, Rubin H, Cooperman BS (1993) J Med Chem 36:3859–3862 45. Zhang Y, An X, Stubbe J, Huang M (2013) J Biol Chem 288:13951–13959 46. Cotruvo JA, Stubbe J (2011) Annu Rev Biochem 80:733–767 47. Crona M, Torrents E, Rohr AK, Hofer A, Furrer E, Tomter AB, Andersson KK, Sahlin M, Sjoberg BM (2011) J Biol Chem 286:33053–33060 48. Cotruvo JA, Stubbe J (2011) Biochemistry 50:1672–1681 49. Cotruvo JA, Stubbe J (2010) Biochemistry 49:1297–1309

123

50. Hammerstad M, Hersleth HP, Tomter AB, Rohr AK, Andersson KK (2013) ACS Chem Biol. doi:10.1021/cb400757h 51. Boal AK, Cotruvo JA, Stubbe J, Rosenzweig AC (2010) Science 329:1526–1530 52. Cotruvo JA, Stich TA, Britt DR, Stubbe J (2013) J Am Chem Soc 135:4027–4039 53. DeLano WL (2002) PyMOL. DeLano Scientific, San Carlos 54. Hanson MA, Schmidt PP, Strand KR, Graslund A, Solomon EI, Andersson KK (1999) J Am Chem Soc 121:6755–6756 55. Tomter AB, Zoppellaro G, Bell CB, Barra AL, Andersen NH, Solomon EI, Andersson KK (2012) PLoS One 7:e33436