Structural Basis for NADH/NAD+ Redox Sensing by a Rex Family Repressor

Molecular Cell

Article Structural Basis for NADH/NAD+ Redox Sensing by a Rex Family Repressor Krystle J. McLaughlin,1,5 Claire M. Strain-Damerell,2,5 Kefang Xie,3,6 Dimitris Brekasis,2 Alexei S. Soares,4 Mark S.B. Paget,2,* and Clara L. Kielkopf1,3,* 1Department

of Biochemistry and Biophysics, University of Rochester School of Medicine and Dentistry, Rochester, NY 14642, USA of Chemistry and Biochemistry, University of Sussex, Falmer, Brighton BN1 9QG, UK 3Department of Biochemistry and Molecular Biology, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD 21205, USA 4Macromolecular Crystallography Research Resource, National Synchrotron Light Source, Brookhaven National Laboratory, Upton, NY 11973, USA 5These authors contributed equally to this work 6Present address: Johns Hopkins School of Medicine, Baltimore, MD 21205, USA *Correspondence: [email protected] (M.S.B.P.), [email protected] (C.L.K.) DOI 10.1016/j.molcel.2010.05.006 2Department

SUMMARY

Nicotinamide adenine dinucleotides have emerged as key signals of the cellular redox state. Yet the structural basis for allosteric gene regulation by the ratio of reduced NADH to oxidized NAD+ is poorly understood. A key sensor among Gram-positive bacteria, Rex represses alternative respiratory gene expression until a limited oxygen supply elevates the intracellular NADH:NAD+ ratio. Here we investigate the molecular mechanism for NADH/NAD+ sensing among Rex family members by determining structures of Thermus aquaticus Rex bound to (1) NAD+, (2) DNA operator, and (3) without ligand. Comparison with the Rex/NADH complex reveals that NADH releases Rex from the DNA site following a 40 closure between the dimeric subunits. Complementary site-directed mutagenesis experiments implicate highly conserved residues in NAD-responsive DNA-binding activity. These rare views of a redox sensor in action establish a means for slight differences in the nicotinamide charge, pucker, and orientation to signal the redox state of the cell.

INTRODUCTION Aerobic life depends on the ability of organisms to detect and respond to environmental oxygen levels. Oxygen deprivation triggers cellular responses that activate anaerobic pathways, increase the efficient use of available oxygen, and activate cellcycle arrest and apoptotic pathways (reviewed in Semenza, 2009; Simon et al., 2008). Conversely, aerobic respiration generates reactive oxygen species, the accumulation of which may damage DNA, lipids, or proteins and contribute to a variety of diseases including cancer, diabetes, cardiovascular disease, and neurodegenerative disorders (Antoniades et al., 2009; Pan

et al., 2009; Rahangdale et al., 2009; Reddy et al., 2009; Zhou et al., 2008). The reoxidation of nicotinamide adenine dinucleotide (NADH) back to NAD+, with a concomitant reduction of oxygen by the electron transport chain, provides the major source of ATP for aerobic organisms (Bohinski, 1987). Since NADH levels become elevated when oxygen is unavailable or the electron transport chain is inhibited (Williamson et al., 1967), the intracellular NADH/NAD+ ratio acts as a sensitive indicator of the redox state. Therefore, in addition to their well-established metabolic functions, pyridine dinucleotides contribute subtle allosteric signals in the regulation of redox-responsive gene transcription (Brekasis and Paget, 2003; Kim et al., 2005; Lamb et al., 2008; Pan et al., 2009; Rutter et al., 2001; Zhang et al., 2002) and related processes including circadian rhythms, apoptosis, and aging (Chen et al., 2009; Lin et al., 2004; Nakahata et al., 2009; Ramsey et al., 2009; Zhao et al., 2008). Structures of eukaryotic proteins that sense pyridine nucleotides have been determined, including CtBP, SIR2, Gal80, and HSCARG (Hawse et al., 2008; Kumar et al., 2008; Nardini et al., 2003; Sanders et al., 2007; Zheng et al., 2007). However, it is not yet clear how subtle chemical differences between oxidized NAD+ and reduced NADH function as signals to regulate gene transcription. Bacteria continually monitor oxygen levels and the cellular redox state by using diverse macromolecular sensors, which are either directly influenced by the oxidation of reactive centers or indirectly influenced by oxygen-sensitive metabolites such as NAD(H) (reviewed in Green and Paget, 2004). Members of the Rex family are widely conserved among Gram-positive bacteria and appear to control respiratory pathways in response to changes in the intracellular NADH/NAD+ redox balance (Figure 1A). In Streptomyces coelicolor, Rex regulates expression of cytochrome bd oxidase (cydABCD), the proton-translocating NADH dehydrogenase from the electron transport chain (nuoAN), heme biosynthetic enzymes, and the Rex gene itself (rexhemACD) (Brekasis and Paget, 2003). Similarly, Bacillus subtilis Rex regulates the cydABCD operon, NADH-linked fermentative lactate dehydrogenase (lctP-ldh), and a putative nitrate transporter (ywcJ) (Gyan et al., 2006; Larsson et al., 2005;

Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc. 563

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Figure 1. Conservation of T-Rex and the Rex Family (A) Schematic diagram of the Rex pathway. Reduced cofactors are colored green, oxidized cofactors are yellow, and Rex is purple. (B) Sequence identity among Rex family members displayed on a representative T-Rex subunit of the T-Rex/NAD+/DNA structure. As indicated in the key, residues are colored by sequence identity as shown in the alignment of Figure S1. Labels indicate the secondary structure elements. The NAD+ is shown as black lines. A view of the WH domain following a 90 rotation about the x axis is given to the right. All residues investigated by site-directed mutagenesis here are labeled and shown in ball-and-stick. (C) Representative SPR sensorgram demonstrating T-Rex association with a consensus ROP, but not a randomized DNA site. A streptavidin sensorchip was charged with the biotinylated oligonucleotide in the presence of excess complementary strand. Sensorgrams from the flow cell with immobilized DNA were corrected by subtracting background values obtained from a DNA-free flow cell, then overlaid. (i) T-Rex (20 nM in duplicate or 5 nM) was injected for 3 min at 20 ml min1 until (ii) the end of injection. (D) Apparent T-Rex and S-Rex affinities for the ROP site and the randomized sequences determined by equilibrium SPR analysis. The average affinities and error bars representing the standard deviations of two independent experiments are shown on the graph. Values of the apparent equilibrium dissociation constants (KD) are approximations due to the high affinity for the ROP site. N.D.B., no detectable binding by a solution containing 10 mM T-Rex protein.

564 Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc.

Schau et al., 2004). Rex homologs from S. coelicolor, B. subtilis, and Thermus aquaticus have been shown to bind either oxidized or reduced pyridine dinucleotides and have a strong preference for reduced NADH association (Brekasis and Paget, 2003; Gyan et al., 2006; Sickmier et al., 2005; Wang et al., 2008). Elevated levels of reduced NADH abolish formation of the Rex complex with the DNA operator (Brekasis and Paget, 2003; Gyan et al., 2006; Wang et al., 2008). In contrast, NAD+ does not inhibit DNA binding but, importantly, competes with NADH for binding to Rex (Brekasis and Paget, 2003). Indeed, the presence of NAD+ has been shown to enhance DNA binding by the B. subtilis Rex homolog (Gyan et al., 2006), and the NAD+ affinity of Rex is enhanced by the presence of DNA (Wang et al., 2008). Therefore, Rex seems to fine-tune gene expression in response to fluctuations in the NADH:NAD+ ratio through the DNA-binding influences of NADH or NAD+ interactions rather than just sensing NADH levels alone. A structural understanding of redox sensing by prokaryotic transcription factors has recently emerged. Structures of the OxyR activation domain in the oxidized and reduced states explain redox-sensitive DNA recognition through the formation of an intramolecular disulfide bond (Choi et al., 2001). More recently, structures of reduced and oxidized OhrR demonstrate that a disulfide bond between the dimeric subunits reorients the winged helix (WH) domains for DNA binding (Hong et al., 2005; Newberry et al., 2007). However, these thiol-mediated redox sensors that stimulate allosteric change by covalent modification are distinct from Rex, which senses the ratio of a dissociable redox couple. The crystal structures of Rex homologs from T. aquaticus (T-Rex), Thermus thermophilus (which is identical in sequence to T-Rex), and B. subtilis (B-Rex) demonstrate a dimeric organization of two modular subunits, each with an N-terminal DNA-binding domain joined to a C-terminal domain for NAD(H) binding (Nakamura et al., 2007; Sickmier et al., 2005; Wang et al., 2008). Previous structures of T-Rex represent the induced state, with two NADH molecules bound near the dimer interface and paired DNA-binding domains arranged in a manner inconsistent with the geometry of a canonical DNA duplex (Nakamura et al., 2007; Sickmier et al., 2005). The NADH/NAD+-sensing mechanism of Rex family members currently remains unknown; however, once understood, it may provide insights for other redox-sensing proteins whose function is regulated by the intracellular NADH/NAD+ ratio. Here we address two key questions: (1) what is the nature of the structural change promoted by NADH that inhibits DNA binding; and (2) how does NAD+ compete with NADH and yet fail to inhibit DNA binding? To address these questions, we first determined the high-resolution structure of T-Rex bound to NAD+ and a consensus Rex operator site (ROP). This structure, together with our previous structure of the induced T-Rex/NADH complex, reveals that NADH association promotes a dramatic conformational change that releases the T-Rex dimer from DNA. Moreover, we identified a mutant ([R90D]T-Rex) that is unable to bind NAD(H) and determined the structures of the (R90D)T-Rex in the absence of ligands or with ROP alone. Together, these structures indicate that T-Rex adopts the DNAbound conformation even in the absence of ligands. Through the use of complementary site-directed mutagenesis, our study has

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

also identified highly conserved residues that are important for NADH/NAD+ sensing. Finally we present a structure-based model to explain the mechanism of induction that is likely to be shared among Rex family members. RESULTS AND DISCUSSION Comparison of T-Rex with Rex Family Members The high degree of phylogenetic sequence conservation (>30% pairwise sequence identities) and excellent model scores (1.0) (see Figure S1, available online, and Figure 1B) suggest common structural mechanisms for redox-dependent gene regulation among T-Rex and more than 30 Rex family members (Sickmier et al., 2005). The T. aquaticus and T. thermophilus Rex homologs (T-Rex) share 100% sequence identity, and accordingly their structures in the NADH-bound states are nearly identical (0.7 A˚ pairwise a-carbon root-mean-square deviation [rmsd] between the Rossmann folds; 2 A˚ rmsd between dimers) (Nakamura et al., 2007; Sickmier et al., 2005). Structural information also exists for the B. subtilis Rex homolog (B-Rex), which shares 53% sequence similarity with T-Rex (Wang et al., 2008). Due to the influence of crystal packing, a symmetry-related B-Rex subunit disrupts the interdomain interactions so that the overall conformation cannot be easily compared with T-Rex (Figure S2A). Nevertheless, the folds of the N- and C-terminal domains (2.9 and 3.4 A˚ rmsd, respectively), dimer interfaces and key residues are maintained at structurally conserved positions of the B-Rex and T-Rex structures (Figures S1 and S2B and S2C). Previously, we demonstrated that reduced NADH dissociates a T-Rex/ROP complex (Sickmier et al., 2005), as established for the mesophilic S. coelicolor (S-)Rex (Brekasis and Paget, 2003) and B. subtilis B-Rex homologs (Gyan et al., 2006; Wang et al., 2008). Oxidized NAD+ competes with NADH for T-Rex binding, and NAD+-bound T-Rex retains DNA-binding activity (Sickmier et al., 2005). Given the strong sequence conservation between T-Rex and other Rex family members, we hypothesized that T-Rex would recognize a similar DNA operator. To test this hypothesis, we used surface plasmon resonance (SPR) to measure T-Rex affinities for 37 base pair double-stranded oligonucleotides comprising either the nuoA-N operator site of S. coelicolor or a randomized DNA sequence. T-Rex binds the S-Rex operator with subnanomolar affinity and lacks detectable binding to a randomized DNA sequence following injection of 10 mM protein (Figure 1C). The S-Rex homolog displays similar DNA sequence specificity and a slightly lower apparent affinity that may be comparable to T-Rex at the physiologically appropriate temperatures for the two homologs (30 C and 65 C, respectively) (Figure 1D). These similarities in primary sequences, structures, NADH/NAD+ sensitivity, and DNA specificity mark T-Rex as a suitable structural model for NADH/ NAD+ sensing by Rex family members. Structure of T-Rex Bound to Oxidized NAD+ and Operator DNA To fully understand the NAD(H)-sensing mechanism of T-Rex and other family members, we first determined the NAD+/DNAbound T-Rex structure (Figure 2A) for comparison with the

NADH-bound structure (Figure 2B). The structure of the T-Rex complex with NAD+ and an oligonucleotide containing a consensus ROP was solved by maximum likelihood molecular replacement methods, and refined at 2.26 A˚ resolution (Table 1). The cocrystallized DNA duplex is a 22 base pair palindrome (50 -CGCTGTGAACGCGTTCACAGCG-30 ), containing two inverted repeats of a ‘‘TGTGAA’’ sequence (bold) that is the minimum palindromic binding site of Rex family members from S. coelicolor and B. subtilis (Brekasis and Paget, 2003; Larsson et al., 2005). The asymmetric unit of the crystal contained a physiologically relevant dimer of T-Rex subunits bound to the double-stranded oligonucleotide, and one molecule of NAD+ (Figure 2A). The bound ROP site maintains a canonical B-DNA shape based on the following observations: a 36 interbase pair twist, a 3.3 A˚ rise, a C20 endo sugar pucker, and no significant curvature of the helical axis. As shown previously for the NADH-bound form, T-Rex is a dimer, and each subunit is composed of an N-terminal WH DNA-binding domain linked to a ‘‘Rossmann’’ fold for binding NAD(H) cofactors. A C-terminal, ‘‘domain-swapped’’ a helix (a8) inserts within a cleft between the WH and the Rossmann domains of the opposite subunit, where it both couples these two domains and mediates the T-Rex dimer. Three residues (residues 58–60) of the flexible glycine-rich ‘‘wing’’ region of the DNA-binding motif are poorly ordered in one of the two subunits, which are otherwise nearly identical within coordinate error (0.3 A˚ rmsd). The final refinement model includes residues 1–206 of one subunit, 1–57 and 61–206 for the other, a single NAD+, and 378 water molecules with a crystallographic R factor of 23% and working R factor of 27%. The compact, ‘‘closed’’ conformation of the NADH-bound structure changed significantly in the DNA bound form (Figure 2C, Movie S1). The previously published structure of T-Rex bound to NADH represents the induced state with two NADH molecules bound near the dimer interface and the paired DNAbinding domains of the molecule positioned in a manner that would clash with canonical DNA duplex (Figure 2B) (Nakamura et al., 2007; Sickmier et al., 2005). In contrast, the complex of the T-Rex dimer with DNA forms a nearly equilateral triangle with molecular dimensions of 70 A˚ per side (Figure 2A). Each subunit of the DNA-bound T-Rex contributes one leg of the triangle, with the C-terminal domains meeting at an angle of 65 at the vertex and the N-terminal domains docked on the DNA operator at the base. The ‘‘recognition’’ helices (residues 41–51) of the T-Rex dimer are related by a 33 A˚ distance (between F43/F430 -Ca atoms) and 179 rotation that fits the consecutive major grooves of the undistorted B-form site. In contrast, the 16 A˚ separation and 170 rotation observed in the T-Rex/NADH complex is incompatible with DNA binding (Sickmier et al., 2005). Release of NADH enables DNA binding by triggering the subunits of the T-Rex dimer to undergo a 43 rigid-body rotation. The hinge of the intersubunit rotation is located near residue D188 within the linker preceding the domain-swapped C-terminal a helix, so that this helix rotates in concert with the other subunit in the dimer. The caliper-like motion between the paired subunits propagates from a rotation of the NAD(H)-binding sites to a dramatic 22 A˚ translation of the WH domains in the dimer. In contrast, very little difference is

Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc. 565

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Figure 2. Structure and Comparison of T-Rex/NAD+/ROP with T-Rex/NADH Complexes (A) X-ray structure of the T-Rex/NAD+/ROP complex. One subunit in the T-Rex dimer is colored slate, the other subunit violet, and NAD+ yellow. The recognition helix (a3) and domainswapped helix (a80 ) are highlighted in brighter shades. The G,C base pairs are colored green, whereas the A,T base pairs are teal. (B) X-ray structure of the T-Rex/NADH complex (PDB ID 1XCB) viewed as in (A) with the NADH in green. A semitransparent DNA surface from the T-Rex/NAD+/ROP structure is shown for reference. Primed italics indicate ligands and residues from a distinct subunit here and throughout. (C) Comparison of the T-Rex/NAD+/ROP structure (blue) with the T-Rex/NADH structure (salmon) following superposition of one of the two Rossmann folds in the dimer. The arrowheads point in the direction of the structural change following NADH association. (Left) View illustrating the movement between the a3 recognition helices of matching subunits to fit the DNA duplex. For clarity, a backbone trace of the T-Rex/NADH subunit used for superposition is semitransparent. The distance is given between F43 Ca atoms at the approximate midpoints of the a3 helices. (Right) View with the foreground subunits now semitransparent traces to expose the movement of the domain-swapped (a80 ) helices with the opposite subunit of the dimer. The distance is given between P203 Ca atoms at the C terminus of the a80 helices. (D) Superposition using all corresponding Ca atoms demonstrates the similarity of the DNAbound (blue) and NADH-bound (salmon) subunits.

observed between the relative orientations of the domains within a given polypeptide chain (Figure 2D). Apart from this global change, the fold of T-Rex in the NAD+/DNA complex remains similar to that of the NADH-bound T-Rex structure, with 2 A˚ rmsd between subunits compared with 6.7 A˚ rmsd between overall dimers. This movement differs from most known repressors, which often (1) locally rotate a DNA-binding domain relative to a regulatory domain rather than the overall subunits and (2) bring a dimeric pair of these DNA-binding domains spatially together rather than apart (Huffman and Brennan, 2002; Lewis, 2005; Pennella and Giedroc, 2005). Interactions between T-Rex and the Operator DNA The core structure and classic mode of DNA recognition via the third, recognition a helix are conserved (Figure 3) (reviewed in Gajiwala and Burley, 2000), despite low pairwise sequence identities between T-Rex and other members of the WH fold family.

566 Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc.

Comparison of the T-Rex WH domain against the Protein Data Bank (http:// www.rcsb.org/pdb) using the DALI server (Holm and Sander, 1993) detected the greatest structural similarity with the diphtheria toxin repressor (DtxR), the founding member of a family of iron-dependent transcription repressors (White et al., 1998) (PDB ID 1DDN; Z score, 6.3; rmsd, 2.4 A˚; sequence identity, 17% for 63 a-carbon pairs). The promoter elements recognized by DtxR are unrelated to the ROP consensus sequence. In addition, the interdomain closure and local helix-to-coil transition that are required for DNA recognition by the DtxR dimer are distinct from the conformational changes of T-Rex (D’Aquino et al., 2005; White et al., 1998). Nevertheless, the spacing and orientation of the WH domains are very similar between the DNA-bound DtxR and T-Rex dimers (Figure S3). Highly conserved residues in the recognition helices of each WH domain contact the major groove edges of the palindromic 50 -TGTGAA half-sites (Figure 3). The interactions are nearly identical between the two T-Rex subunits. Residues Y51 and the hydrophobic portion of K47 indirectly recognize the terminal adenines (8 and 9) of the ROP site by packing against the methyl

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Table 1. Summary of Data Collection and Refinement Statistics Crystallographic Data T-Rex/NAD+/DNA Operator

(R90D)T-Rex/ DNA Operator

Apo-(R90D)T-Rex

Resolution range (A˚)

20.02.26

20.02.49

20.02.40

Space group Unit cell (A˚)

P3121 a = b = 63.15, c = 299.26

a = b = 62.95, c = 298.11

a = 57.00, b = 88.22, c = 99.58

Redundancy

10.7 (11.5)

7.0 (6.5)

7.1 (3.3)

P212121

Completeness (%)

97.6 (99.8)

95.2 (90.1)

98.5 (92.2)

Rsym (%)

8.6 (40.1)

6.2 (48.6)

14.3 (25.1)



21.3 (7.2)

17.9 (3.2)

13.9 (1.9)

23.4/26.8

25.4/28.5

21.8/25.6

Refinement Statistics Rcryst/Rfree (%) Number of Nonhydrogen Atoms Protein

3180

3176

3178

DNA

896

896

–

NAD+ (C21H27N7O14P2)

44

–

–

Water

378

122

217

Bond lengths (A˚)

0.007

0.014

0.008

Bond angles ( )

1.48

1.57

1.53

Rmsd

Average B Factors (A˚2) Overall

44.3

52.6

25.4

Protein

42.3

53.1

24.5

DNA

48.9

49.1

–

NAD+

45.5

–

–

Overall G factor

0.34

0.28

0.34

Ramachandran Analysis (%) Most favored

92.1

90.1

91.5

Additionally allowed

7.6

9.4

8.2

Generously allowed

0.3

0.6

0.3

Values in parentheses are for the highest-resolution shell, 2.37–2.26 A˚ for T-Rex/NAD+/DNA data set, 2.59–2.49 A˚ for (R90D)T-Rex, and 2.49–2.40 A˚ for P P P P apo(R90D)T-Rex. An I/s or resolution cutoff was not applied during structure refinement. Rsym= hkl ijIi  j/ hkl i Ii where Ii is an intensity I for the P ith measurement of a reflection with indices hkl and is the weighted mean of all measurements of I. Rcryst = hklkFobs(hkl)j  kjFcalc(hkl)k)/ P hkljFobs(hkl)j for the working set of reflections, Rfree is Rcryst for 7.5% of the reflections excluded from the refinement. In agreement with the weak electron density, the average B factor of a second NAD+ was 145 A˚2 (200 A˚2 maximum B factor). This low-occupancy NAD+ was not included in PDB 3IKT. The overall G factor is a measure of a structure’s ‘‘normalness’’ on a scale of 1.0 to 1.0, where negative values may be problematic. The G factors of T-Rex/NAD+/ROP, apo-(R90)T-Rex, and (R90)T-Rex/ROP are significantly better than typical values of 0.9 to 0.2 for 2.3 A˚ and 0.8 to 0.3 for 2.4–2.5 A˚ resolution (Laskowski et al., 1993).

groups of the base-paired thymines (140 and 150 ) (primed italics indicate residues or nucleotides of the opposite subunit) (Figure 3A). An F43 side chain packs against the hydrophobic edge of cytosine16 in the duplex strand (Figure 3B). However, F43 may contribute to the higher ROP affinity observed for T-Rex, since this residue differs from the S-Rex and B-Rex homologs. The K47 side chain interacts with the guanine7 base through a bifurcated hydrogen bond with the exocyclic O6 and imino N7 positions (Figure 3B). A neighboring R46 residue recognizes the O6 and N7 positions of guanine5 by donating two hydrogen bonds in a manner considered prototypical for arginine residues (Seeman et al., 1976) (Figure 3C). Changes of this invariant guanine5 severely inhibit DNA binding by B-Rex (Wang et al., 2008). Site-directed mutagenesis was used to

further explore the significance of the R46 and K47 contacts with the guanines in the ROP site. Alanine substitutions were introduced into the S-Rex protein at either the R46 or K47 S-Rex counterparts (R59 or K60). These mutations completely abolished detectable binding of the variants to the S. coelicolor rex gene ROP site at protein concentrations up to 1 mM (Figure S4). This finding highlights the importance and conservation of the recognition helix contacts among Rex family members. Adjacent to the core recognition helix, a strictly conserved, glycine-rich b strand/loop or wing region (residues 53–62) provides additional interactions with the minor groove (Figure 3D). In both T-Rex subunits, two glycines and an intervening tyrosine (G53, G56, and Y55) cross the phosphodiester backbone and enter the minor groove as expected. Residues G61

Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc. 567

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Figure 3. Interactions of T-Rex and the Rex Operator DNA (A–C) The recognition helix of a representative T-Rex subunit contacting the ROP site. (A) is viewed in a similar orientation to Figure 2A; (B) and (C) are slabbed down the DNA axis from back to front. The carbons of the DNA are colored by sequence as in Figure 2A. Water molecules are blue spheres. Hydrophobic contacts described in the text are shown as semitransparent surfaces. Distances of putative hydrogen bonds are labeled in A˚ units. The T-Rex R46 and K47 counterparts of residues investigated by site-directed mutagenesis are underlined and the carbons colored magenta. (D) Detailed view of the wing region of one T-Rex subunit contacting the DNA oligonucleotide viewed in a similar orientation as (A). Two alternative conformations of R58 are designated ‘‘a’’ or ‘‘b.’’ (E) Schematic diagram of contacts between one of the T-Rex subunits and the unique half-site of the palindromic DNA duplex. The pseudodyad axis relating the left and right halves of the DNA is indicated by an oval. All DNA interactions by the second T-Rex subunit are identical to those shown for the first, with the exception that R58 and G59 are not observed in the electron density. The ROP consensus is emphasized by bold font; primed italic font distinguishes one of the two DNA strands. Filled lines represent interactions mediated by peptide bonds of the backbone, whereas dashed lines represent interactions with Ca or side-chain atoms. Arrowheads are used to mark hydrogen bonds or ionic interactions, whereas filled circular ends mark hydrophobic contacts. Filled blue circles represent water-mediated contacts. Residues in the recognition helix are shown in dark red font, residues in the wing region are filled dark blue with white font, and residues tested by site-directed mutagenesis are underlined and shaded.

and Y62 return the polypeptide to the major groove (Figure 3E). However, the central portions of the wings differ between the two subunits of the T-Rex dimer. In the loop apex of one of the two subunits, the side chain of R58 exhibited two alternative conformations in the electron density (labeled R58a and R58b). Both of these and an adjacent glycine (G59) insert within the minor groove (Figures 3D and 3E). In the second subunit, these residues are disordered and lack jFoj  jFcj electron density at 2 s contour level. The distinct crystal-packing environments of the DNA termini may contribute to the subunit differences, since the well-ordered T-Rex wing is found within the narrower of the two minor grooves (a minor groove width of 3–5 A˚ for

568 Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc.

cytosine3/guanine200 versus 6–8 A˚ for cytosine30 /guanine20). A similar observation has been made in the prototypical DtxR repressor structure: two out of four crystallographically independent wings of the DtxR structure insert arginine residues deep within a narrow minor groove, whereas arginine residues of the remaining two wings make peripheral contacts with the DNA backbone (White et al., 1998). A Single Bound NAD+ per T-Rex Dimer One striking feature of the T-Rex/NAD+/ROP complex is the presence of a single, well-defined NAD+ molecule per dimer (Figure 4). The omit electron density for this ‘‘major’’ NAD+ is

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Figure 4. Oxidized NAD+ Bound to the T-Rex/ROP Complex (A) Composite omit electron density at the 1s contour level for the high-occupancy (major) NAD+ and low-occupancy (minor) NAD+0 -binding sites of the T-Rex/ NAD+/ROP structure. The NAD+0 coordinates shown were not included in the final coordinates due to the high temperature factors following refinement. (B) Fluorescence spectra of washed T-Rex/NAD+/ROP crystals following excitation at 340 nm demonstrate the lack of NADH emission at the characteristic 425 nm wavelength (dashed line), compared with a standard solution containing a 1:1 molar ratio of purified T-Rex protein and ROP oligonucleotide. The spectra were scaled by emission at 340 nm following excitation of tryptophan fluorescence at 285 nm. AU, arbitrary units. (C) View into the Rossmann folds of the T-Rex/NAD+/ROP structure. Expanded view on the right illustrates the distance between nicotinamide C5 atoms of the major NAD+ (yellow) and the minor NAD+0 (gray). The asymmetric F189 of the T-Rex/NADH complex is shown for comparison (red). (D) View rotated 90 about the x axis relative to (C). Expanded view shows stick representations of the NAD+ (yellow) and interacting residues, with distances given in A˚. For contrast, the corresponding view of T-Rex bound to NADH (green) (PDB ID 1XCB) is shown to the far right. (E) (Left) Observed structures NAD+ (yellow) and NADH (green) bound to T-Rex. (Right) Models show the poor fit of the anti NADH (green) with the polypeptide of DNA-bound T-Rex (magenta/blue subunits), and the syn NAD+ (yellow) overlap with the polypeptide of the NADH-bound T-Rex (salmon/pink subunits). The dipole moment of the a helix is represented by 40. Models were generated by superposition of the NAD(H)-binding domains.

Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc. 569

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

consistent with the planar aromatic ring of an oxidized nicotinamide, whereas little electron density is observed in the second, ‘‘minor’’ site of the dimer (Figure 4A). Fluorescence spectra of washed T-Rex/NAD+/ROP crystals further confirm that little or no NADH is tightly bound within the crystallized complex (Figure 4B). The different occupancies of the paired NAD+ sites do not rely on distinct conformations of the two subunits of the dimer, since the conformations of these subunits are nearly indistinguishable in the DNA-bound form (0.3 A˚ rmsd). Instead, the relatively close distance predicted for a pair of bound NAD+ molecules (7 A˚) is most likely the reason for the presence of a single ‘‘major’’ NAD+-binding site (Figure 4C). The proximity would lead to a strong electrostatic repulsion between the positive charges of the two nicotinamide rings (5 A˚ spatial extension per ring with a 10 A˚ assumption for two rings) (Guillot et al., 2003). In both states, the NAD+ and NADH dinucleotide(s) bind at the dimer interface, with the adenosine portion engaged by the P loop of one Rossmann fold and the nicotinamide group by the N terminus of the domain-swapped a helix and the a5/b4 loop of the opposing subunit (Figure 4D). However, two key changes in the environment of the nicotinamide are expected to enhance the electrostatic repulsion between two bound NAD+ molecules. First, an asymmetric F189 of the NADH-bound T-Rex that inserts between the two molecules of bound NADH has rearranged in the DNA-bound form (Figure 4C). This rearrangement allows the F189 residue of the DNA-bound conformation to pack within the hydrophobic interior of the opposite subunit and prevents it from insulating adjacent NAD+-positive charges with its hydrophobic side chain. Comparison of three crystallographically independent dimers of T-Rex/NADH (Sickmier et al., 2005) and the T. thermophilus T-Rex/NADH structure (Nakamura et al., 2007) confirms that the asymmetric position of F189 between the NADH molecules correlates with 7 –15 local rotation of the WH domain of the neighboring subunit away from the DNAbound conformation (Figure S5). Therefore, the overall fit of the T-Rex dimers with the DNA grooves improves after the F189 side chain is repositioned in the T-Rex/NAD+/ROP complex. Second, the domain-swapped a helix is expected to contribute to the electrostatic repulsion due to an estimated +0.7 charge from the a-helical dipole moment (Hol, 1985). In the DNA-bound form, the 43 rotation of this domain with the opposing subunit is likely to further enhance the unfavorable electrostatic field by reorienting the helix dipole axes directly toward the former location of the NADH carboxamide in the anti conformation. The influence of these two structural changes, together with the established electrostatic repulsion between two positively charged NAD+ molecules (Guillot et al., 2003), offers a straightforward explanation for the observation of a single, well-defined NAD+ molecule per T-Rex dimer.

ular hydrogen bond with the adjacent phosphoryl group of the dinucleotide. These interactions differ from the direct hydrogen bonds previously observed between the reduced, anti nicotinamide and the residues preceding the domain-swapped a helix. In all structures of T-Rex/NADH, the carbonyl group of V187 forms hydrogen bonds with the carboxamide nitrogen of the anti NADH (Nakamura et al., 2007; Sickmier et al., 2005). An additional hydrogen bond between the anti NADH carboxamide oxygen and the F189 peptide nitrogen is observed only for the subunits in which F189 inserts between the reduced nicotinamide groups (Nakamura et al., 2007; Sickmier et al., 2005). Although the syn/anti preferences of free NAD+ compared to NADH are considered to be slight (less than 10-fold) (Oppenhiemer and Handlon, 1992), analogous syn to anti transitions of bound NAD(H) have been observed among several pro-Sstereospecific enzymes, such as transhydrogenase (Prasad et al., 2002) and UDP-galactose 4-epimerase (Thoden et al., 1996). Three structural changes in the T-Rex subunits appear to promote the oxidation-dependent switch between syn and anti conformers (Figures 4D and 4E): (1) steric overlap, (2) loss of a hydrogen bond, and (3) orientation of the a-helical dipole moment. First, the position of F189 in the DNA-bound form is expected to clash with that of the carboxamide group of the anti NADH cofactor due to rotation of the domain-swapped helix with the opposite subunit in the dimer (2.4 A˚ Cb—NO7; 2.8 A˚ Cd1—NC3) (Figure 4E). Conversely, the syn position of the NAD+ carboxamide group is expected to severely clash with the NADH-bound position of the A94 backbone and side chain (1.5 A˚ Cb—NO7). These steric overlaps would prohibit syn NADH cofactors from binding the induced T-Rex state and vice versa. Second, the rearrangement of F189 to the DNAbound conformation breaks a hydrogen bond that previously secured one of the two nicotinamides of the asymmetric NADHbound dimer in the anti position (Figure 4D, Figure S5). Third, the carboxamide nitrogen of the anti NADH is located directly adjacent to the N-terminal turn of the domain-swapped a helix (Figure 4E). In this position, an oxidized NAD+ state is expected to be unfavorable, given the partial positive charge of the a-helical dipole moment (Hol, 1985). Accordingly, transition to the syn conformer in the T-Rex/NAD+/ROP complex achieves a greater than 6 A˚ separation between these positively charged regions. A similar electrostatic phenomenon may contribute to the epimerase structures, where the N-terminal turn of an a helix moves from less than 3.8 A˚ from the carboxamide group of anti NADH to greater than 10 A˚ from the syn NAD+ (Thoden et al., 1996; our unpublished data). Similarly, an electrostatic repulsion between NAD+ and a nearby arginine is proposed to influence the syn NAD+/anti NADH preferences of transhydrogenase (Prasad et al., 2002).

Syn Conformation of the Bound NAD+ The oxidized nicotinamide ring of the bound NAD+ molecule adopts a syn conformation, which has rotated 180 about the glycosyl torsion angle relative to the anti conformation of bound NADH (Figure 4D). The carboxamide nitrogen of the syn NAD+ is favorably engaged by a water-mediated hydrogen bond with the A94 carbonyl oxygen of T-Rex. In addition, it has an intramolec-

(R90D)T-Rex Matches the DNA-Bound Conformation in the Absence of NAD(H) Cofactors To investigate whether the binding to DNA or to NAD+ mediates the structural changes observed in T-Rex/NAD+/ROP complex, we determined structures of T-Rex in the absence of NAD(H), bound or free from DNA (Figure 5, Table 1). Since wild-type T-Rex protein copurifies with bound NADH, a mutation in the

570 Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc.

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Figure 5. Structures of (R90D)T-Rex in the Absence of Cofactors Match the DNA-Bound Form (A) Absorption spectra demonstrating that (R90D)T-Rex does not copurify with NADH. Purified T-Rex proteins (i) before (light gray) and (ii) after (gray) depletion of NADH by acidic ammonium sulfate precipitation are compared with (iii) (R90D)T-Rex (black). The absorbance at 340 nm is characteristic of reduced NADH. (B) SPR sensorgrams as described for Figure 1A demonstrating the NADH concentration dependence of T-Rex dissociation from the ROP site, and that NAD+ does not dissociate the T-Rex/ROP complex. (C) SPR sensorgrams indicating that NADH concentrations greater than 50 nM are required to dissociate (R90D)T-Rex from the ROP site. (i) (R90D)T-Rex was injected to give a response unit shift of 200 RU, (ii) end of T-Rex injection, (iii) injection of either 50 nM or 1 mM NADH. (D) (R90D)T-Rex/ROP (cyan) with T-Rex/NAD+/ROP (red) complexes following superposition of corresponding Ca atoms. (E) Structural comparison of apo(R90D)T-Rex (blue) with T-Rex/NAD+/ROP (red) following superposition of corresponding Ca atoms.

P loop of the dinucleotide-binding domain (R90D) was used to facilitate NADH removal. Absorbance spectroscopy confirmed that the purified (R90D)T-Rex protein lacked detectable NADH (Figure 5A). The concentration of NADH required for DNA dissociation was markedly higher for the (R90D)T-Rex mutant than the wild-type protein (Figures 5B and 5C). The structure of (R90D)TRex was determined in complex with the 22 bp ROP oligonucleotide and refined to 2.5 A˚ resolution. The overall structure of the T-Rex/ROP complex was indistinguishable within coordinate error from that of the wild-type T-Rex/NAD+/ROP complex (0.3 A˚ rmsd) (Figure 5D). Therefore, the T-Rex/ROP complex presents the NAD+-binding site without the need for significant structural rearrangements. In support of these findings, although NAD+ is not required for DNA binding by Rex family members (Brekasis and Paget, 2003), the presence of NAD+ enhances DNA association (Gyan et al., 2006; Wang et al., 2008). These data are consistent with the hypothesis that NAD+ binding allows Rex to primarily function as a sensor of NAD+/NADH ratios rather than as a simple monitor of NADH concentration. To reveal any possible DNA-induced effects, we determined the apo-(R90D)T-Rex structure and refined to 2.4 A˚ resolution (Table 1). The overall structure of apo-(R90D)T-Rex closely matches the DNA-bound conformation even in the absence of DNA or NAD+ ligands (0.8 A˚ rmsd) (Figure 5E). Neither crystal contacts nor the R90D mutation appear to influence the overall conformations. The apo-(R90D) and DNA-bound T-Rex crystal-

lize in distinct space groups that do not apparently restrict the interdomain or intersubunit orientations. The R90—D970 and R900 —D97 intersubunit salt bridges observed for the wild-type T-Rex (Figure 4D) are replaced by repulsive electronegative charges in the R90D mutant. Nevertheless, the R90D and D970 backbone Ca atoms move 7 A˚ closer in the apo-(R90D)T-Rex structure than in the NADH-bound conformation, indicating that the match with the DNA-bound state is not likely to be influenced by the mutation. This characteristic of the apo state adopting the DNA-bound conformation also has been observed for the disulfide-regulated repressor OhrR (Hong et al., 2005; Newberry et al., 2007), as well other repressors including HucR and MexR (Bordelon et al., 2006; Lim et al., 2002). These observations suggest that the DNA-bound state is a major conformation of Rex in the absence of ligands. Significance of Y98—D1880 to R16—D1880 Exchange for NADH/NAD+ Sensing Following release of NADH, the rotation between the T-Rex subunits causes the interactions of Y98—D1880 to exchange for R16—D1880 (Figure 6A, Movie S2). The D1880 residue lies on the axis of rotation preceding the domain-swapped C-terminal helix, and it remains relatively stationary when the structures are compared. In contrast, Y98 in the Rossmann fold and R16 in the WH domain undergo large movements following the intersubunit rotation. Due to a shift of 4 A˚ (Ca-Ca

Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc. 571

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Figure 6. Residues for NADH/NAD+ Sensing by Rex Family Proteins (A) View of the R16, D1880 , Y98 triad and flanking P1730 and I12 residues from the T-Rex/NAD+/ ROP structure (left) compared with the T-Rex/ NADH structure (right, PDB ID 1XCB). (B) S-Rex mutations, corresponding T-Rex residues, and properties of the mutant proteins. The a-helical content was estimated from circular dichroism spectra shown in Figure S7 using the formula of Greenfield and Fasman (1969): % a helix = ([q]208nm – 4,000)/(33,000 – 4,000). For comparison, 44% of the residues in the crystal structure adopt f/c angles of an a helix. The ability to bind NADH was established by absorption spectra of purified proteins (Figure S8A). Although (D203S)S-Rex did not copurify with detectable NADH, it weakly bound NADH by isothermal titration calorimetry (Figure S8B). A reliable estimate of the affinity could not be obtained since the binding reaction reached equilibrium very slowly (>7 min). The IC50 of NADH for dissociation of the (WT)SRex/ROP and (Y111F)S-Rex/ROP complexes was determined by SPR as shown in Figure S8C. The average IC50 values and standard deviations from two independent experiments are given. (C) Model for NADH/NAD+ sensing by Rex family members.

distances), the Y98 side chain continues to shield the nicotinamide ring from the solvent despite the anti-to-syn transition between the reduced and oxidized states. Nevertheless, the conformational changes disrupt an interaction between the Y98 hydroxyl group and the D1880 side chain. Simultaneously, the R16 residue of the WH domain has moved up to 6 A˚ (NH1NH1 distances) to form a salt bridge with the D1880 side chain. This R16 movement is enhanced by the 7 local rotation of the WH domain, which is correlated with the relocation of the asymmetric F189 (Figure S5). To form the salt bridge with the D1880 in the DNA-bound state, an extended R16 conformer is captured between the hydrophobic side chains of a neighboring I12 and P1730 from the Rossmann fold of the opposite subunit (Figure 6A, Figure S6). In contrast, the R16 residues of the seven crystallographically independent, NADH-bound

572 Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc.

T-Rex subunits are solvent exposed and located 9 A˚ and 16 A˚ from D1880 and P1730 , respectively. Based on these observations, we hypothesized that the salt bridge between the highly conserved R16 and D1880 residues may lock the WH domains of Rex family members for DNA binding. Conversely, the interaction between the D1880 side chain and Y98 might secure the NADH-bound conformation. To test this hypothesis, we mutated the corresponding residues of S-Rex (summarized in Figure 6B). The similar secondary structure contents of the mutant and wild-type proteins were verified by circular dichroism spectroscopy (Figure S7). The capacities of the mutant and wild-type proteins to bind NADH were evaluated by the absorbance at 340 nm following size exclusion chromatography (Figure S8), and their abilities to bind a promoter fragment were assessed by electrophoretic mobility shift assay (EMSA) (Figure S4). To test the importance of the R16—D1880 salt bridge for the DNA-bound conformation, we mutated the corresponding S-Rex residues (R29 and D203) to alanine and serine, respectively (a D203A mutant was poorly expressed). Remarkably, the (R29A)S-Rex mutant lacked detectable association with a ROP fragment at protein concentrations as high as 0.25 mM (greater than 1000-fold penalty). Since this residue is more than 20 A˚ from the DNA-binding site, the inhibitory effects of this mutation are most likely due to an indirect influence on

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Rex protein conformation. In support of a dual role for D188, the purified (D203S)S-Rex mutant both lacked detectable copurified NADH (in contrast to R29A; Figures S7 and S8) and failed to interact with ROP DNA (Figure S4). Together, these results support the hypothesis that the R16—D1880 salt bridge is required to lock the positions of the WH domains within the dimer for high-affinity DNA binding by Rex family proteins. We investigated a possible role for the D1880 —Y98 interaction of the NADH-bound state to communicate the NAD(H) redox state to the WH domains. Since Rex appears to require the hydrophobic surface area of the Y98 side chain for high-affinity interactions with NAD(H), the following mutations were introduced in place of the Y98 counterpart: (1) a phenylalanine that lacks the hydroxyl group for engaging the D188 counterpart but retains an aromatic ring for nicotinamide interactions (Y111F), and (2) an arginine expected to compete with the counterpart of R16 for the D188 salt bridge (Y111R). The (Y111F)SRex protein bound the ROP-containing DNA with comparable affinity to wild-type (WT)S-Rex (Figure S4) and retained the ability to bind NADH (Figure S8). Since favorable interactions are expected between the quadrupole moment of a phenylalanine side chain and a negatively charged aspartate (Dougherty, 1996), we hypothesized that (Y111F)S-Rex would be slightly less responsive to NADH than (WT)S-Rex. As expected, the concentration of NADH required to dissociate (Y111F)S-Rex from ROP DNA was 3-fold greater than observed for (WT)SRex (IC50 2.5 mM for [Y111F]S-Rex versus 0.8 mM for [WT]SRex) (Figure S8C). For (Y111R)S-Rex, we hypothesized that the positively charged arginine side chain would form a salt bridge with D1880 , thereby preventing formation of the R16—D1880 salt bridge and hence the DNA-bound state. Accordingly, the (Y111R)S-Rex lacked detectable DNA binding (Figure S4), suggesting that this variant could not achieve the DNA-bound conformation. Altogether, these data are consistent with a dual role for D1880 in stabilizing both the DNA-bound and NADHbound conformations through its alternative interactions with R16 and Y98, respectively. Prior T-Rex/NADH structures suggested the F189 residue plays a key role in changing the relative orientations of the Rossmann and WH domains for redox-sensitive DNA binding (Sickmier et al., 2005) (Figure S5). Our data now demonstrate that a primarily rigid-body, intersubunit rotation occurs following NADH dissociation. An S-Rex mutant with the F189 counterpart (L204) replaced with alanine was unable to bind NADH (data not shown). This observation is consistent with the F189 location between the nicotinamide rings of the NADH-bound T-Rex (Figure 4C, Figure S5). As such, it remains possible that F189 may play a role in adjusting the interdomain orientation in response to NADH. Conclusions and Outlook The structures of T-Rex/NAD+/ROP and T-Rex/NADH reveal that NADH binding to T-Rex causes the subunits to rotate so that the dimer can no longer bind to DNA. In contrast, the (R90D)T-Rex/ ROP compared with the T-Rex/NAD+/ROP structures demonstrate that NAD+ binding does not require the DNA-bound state of T-Rex to change. Additionally, the structure of apo-(R90D)TRex in the absence of ligand shares a nearly identical conforma-

tion with T-Rex bound to DNA. Based on these results, we propose the following molecular pathway for NADH/NAD+ sensing, starting from the DNA-bound state (Figure 6C): (1) syn NAD+ dissociation allows NADH to bind, and the subunits rotate to optimize interactions with anti NADH; (2) the Y98 hydroxyl group engages D1880 , which in turn weakens and uncouples the salt bridge between D1880 and R16 from the WH domain of the paired subunit; and (3) Rex is released from the DNA due to a significant overlap between the WH domains and the DNA site. In parallel, F189 inserts between the reduced nicotinamide groups, so that hydrogen bonds between that F189 backbone and the anti NADH cofactor induce a slight rotation of the WH domains that further weakens the R16—D1880 salt bridge. The Y98, D188, and R16 residues as well as the hydrophobic nature of residue F189 are highly conserved among Rex family members (Figure 1B, Figure S1). Therefore, the NADH-, apo-, and NAD+/DNA-bound T-Rex structures are expected to represent the major conformational stages of redox signaling across members of the Rex family. Beyond this prokaryotic pathway, several eukaryotic dinucleotide-binding proteins have been suggested to link the redox state of the NAD(P)H/NAD(P)+ pool to cellular responses. Among these, similarities are revealed by comparison of the T-Rex/ NAD+/ROP structure with that of the nitric-oxide regulator HSCARG bound to NADP+ (Zheng et al., 2007) (Figure S9). These include a single syn NAD(P)+ cofactor per dimer, a tyrosine residue shielding the external face of the nicotinamide ring, and appropriately located backbone atoms for interactions with an anti conformer of a reduced cofactor. Despite these tantalizing hints, whether the structural changes described here for T-Rex are common themes among prokaryotes and eukaryotes await further structural portraits of NADH/NAD+ sensors in the oxidized and reduced states. EXPERIMENTAL PROCEDURES Protein and DNA Preparation, Crystallization, and Structure Determination Full-length wild-type T-Rex and a mutant (R90D)T-Rex protein, in which R90 in the P loop of the Rossmann fold was replaced by an aspartate, were overexpressed in BL21(DE3). Each T-Rex protein was purified by Ni2+-affinity chromatography, followed by protease cleavage of the 6xHis tag and cation exchange chromatography. To improve solubility and decrease proteolytic susceptibility, the wild-type and mutant S-Rex proteins included residues 5–233, which lacks a C-terminal region that is predicted to be intrinsically unstructured. S-Rex proteins and variants expressed using pET15b in BL21(DE3)pLysS were purified by Ni2+-affinity chromatography followed by size exclusion chromatography as described (Brekasis and Paget, 2003). PAGE-purified synthetic DNA oligonucleotides were purchased from IDT, Inc. The cocrystallized complexes contained 0.25 mM T-Rex polypeptide and an excess of 0.75 mM DNA oligonucleotide. Complexes with palindromic blunt-ended DNA oligonucleotides of various lengths (18–24 bp in 2 bp increments) all produced crystals under similar conditions; however, only the 22 bp oligonucleotide of sequence 50 - CGCTGTGAACGCGTTCACAGCG-30 (ROP consensus site in bold) yielded diffraction data of sufficient quality for structure determination. A final concentration of 1 mM NAD+ (pH 7.0) was included during cocrystallization of the T-Rex/NAD+/ROP complex. The structures were determined using molecular replacement methods. Details of the crystallization, data collection, structure solution, and refinement are described in the Supplemental Experimental Procedures, and the statistics are presented in Table 1.

Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc. 573

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Surface Plasmon Resonance Experiments SPR experiments described in Figures 1 and 5 used HBS-T running buffer (10 mM HEPES [pH 7.4], 150 mM NaCl, 3.4 mM EDTA, 0.005% v/v Tween-20) at a flow rate of 20 ml min1 and temperature of 25 C with a BIAcore instrument (GE Healthcare). The 50 biotinylated oligonucleotide (ROP, 50 -Bi-AGATCGCG AACATGTGAAGCAGGTCACAAGCCCAACT-30 ; Random, 50 -Bi-AGATCGC GAACACCTCATAGACTATTGAAGCCCAACT-30 ) was annealed with a 1:4 excess of the nonbiotinylated oligonucleotide (ROPc, 50 -GGTTGGGCTTGT GACCTGCTTCACATGTTCGCGATCT-30 ; RandomC, 50 GGTTGGGCTTCAAT AGTCTATGAGGTGTTCGCGATCT-30 ), and 300 RU of the DNA duplex was immobilized. To avoid mass transfer effects, equilibrium analysis was used to estimate the DNA affinities as described in Sickmier et al. (2006) with extended injection times. Sensorchip surfaces were regenerated by injection of 1 M NaCl. SPR experiments to measure the NADH response of (Y111F)TRex are described in the Supplemental Experimental Procedures and shown in Figure S8C.

Fluorescence Measurements of NADH Content Crystals harvested from hanging drops were washed three times in 200 ml of crystallization solution and resuspended in 500 ml water for fluorescence experiments. A standard solution composed of 1.74 mM purified T-Rex protein and 1.74 mM crystallized DNA was prepared for comparison. Fluorescence measurements were made using a SPEX Fluoromax-3 connected to a circulating water bath at 22 C. Blank scans of water between samples were subtracted to correct the experimental data. No fluorescence was detected from the buffer used for crystallization. Spectra were scaled for differences in protein concentration by differences in tryptophan emission intensity. Tryptophan fluorescence was excited at 285 nm and the emission spectra monitored at 340 nm. Although Fo¨rster resonance energy transfer (FRET) to bound NADH may in principle decrease the apparent emission intensity from tryptophan, FRET is unlikely to affect the comparison here for several reasons: (1) no maxima were observed in the 425 nm region of the emission spectra following excitation at 285 nm, (2) no tryptophans are located close to the nicotinamide rings of the bound NADH (closest distance 20 A˚ W81-NADH, compared with 25 A˚ Fo¨rster radius), and (3) any decreases in apparent concentration due to FRET are expected to affect both the standard solution of protein/DNA complex and dissolved crystals in a comparable manner. NADH was excited directly at 340 nm and emission spectra collected from 400 to 500 nm.

ACCESSION NUMBERS Coordinates are available from the RCSB Protein Data Bank under the following ID codes: T-Rex/NAD+/ROP, 3IKT; apo-(R90D)T-Rex, 3IKV; and (R90D)T-Rex/ROP, 3IL2.

SUPPLEMENTAL INFORMATION The Supplemental Information includes Supplemental Experimental Procedures, two movies, and nine figures and can be found with this article online at doi:10.1016/j.molcel.2010.05.006.

ACKNOWLEDGMENTS We deeply appreciate S.K. Burley’s assistance developing this project. We are grateful to Y. Chen for technical assistance, Y. Lin and E. Sickmier for preliminary crystallographic analysis, J. Wedekind and C. Brooks for comments on the manuscript, and J. Kielkopf and D. Damerell for guidance with calculations. We thank C. von Wachenfeldt for sharing preliminary results prior to publication. The National Synchrotron Light Source is supported by the U.S. Department of Energy and the National Institutes of Health. This work was supported by a Lang Award to C.L.K., Biotechnology and Biological Sciences Research Council (BBSRC) grant P19928 to M.S.B.P, and a BBSRC studentship to C.M.S.-D.

574 Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc.

Received: October 9, 2009 Revised: February 21, 2010 Accepted: May 11, 2010 Published: May 27, 2010 REFERENCES Antoniades, C., Antonopoulos, A.S., Bendall, J.K., and Channon, K.M. (2009). Targeting redox signaling in the vascular wall: from basic science to clinical practice. Curr. Pharm. Des. 15, 329–342. Bohinski, R.C. (1987). Oxidative phosphorylation. In Modern Concepts in Biochemistry (Newton, MA: Bacon, Inc.), pp. 567–604. Bordelon, T., Wilkinson, S.P., Grove, A., and Newcomer, M.E. (2006). The crystal structure of the transcriptional regulator HucR from Deinococcus radiodurans reveals a repressor preconfigured for DNA binding. J. Mol. Biol. 360, 168–177. Brekasis, D., and Paget, M.S.B. (2003). A novel sensor of NADH/NAD+ redox poise in Streptomyces coelicolor A3(2). EMBO J. 22, 4856–4865. Chen, S., Whetstine, J.R., Ghosh, S., Hanover, J.A., Gali, R.R., Grosu, P., and Shi, Y. (2009). The conserved NAD(H)-dependent corepressor CTBP-1 regulates Caenorhabditis elegans life span. Proc. Natl. Acad. Sci. USA 106, 1496–1501. Choi, H., Kim, S., Mukhopadhyay, P., Cho, S., Woo, J., Storz, G., and Ryu, S. (2001). Structural basis of the redox switch in the OxyR transcription factor. Cell 105, 103–113. D’Aquino, J.A., Tetenbaum-Novatt, J., White, A., Berkovitch, F., and Ringe, D. (2005). Mechanism of metal ion activation of the diphtheria toxin repressor DtxR. Proc. Natl. Acad. Sci. USA 102, 18408–18413. Dougherty, D.A. (1996). Cation-pi interactions in chemistry and biology: a new view of benzene, Phe, Tyr, and Trp. Science 271, 163–168. Gajiwala, K.S., and Burley, S.K. (2000). Winged helix proteins. Curr. Opin. Struct. Biol. 10, 110–116. Green, J., and Paget, M.S. (2004). Bacterial redox sensors. Nat. Rev. Microbiol. 2, 954–966. Greenfield, N., and Fasman, G.D. (1969). Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry 8, 4108–4116. Guillot, B., Muzet, N., Artacho, E., Lecomte, C., and Jelsch, C. (2003). Experimental and theoretical electron density studies in large molecules: NAD+, b-nicotinamide adenine dinucleotide. J. Phys. Chem. B 107, 9109–9121. Gyan, S., Shiohira, Y., Sato, I., Takeuchi, M., and Sato, T. (2006). Regulatory loop between redox sensing of the NADH/NAD+ ratio by Rex (YdiH) and oxidation of NADH by NADH dehydrogenase Ndh in Bacillus subtilis. J. Bacteriol. 188, 7062–7071. Hawse, W.F., Hoff, K.G., Fatkins, D.G., Daines, A., Zubkova, O.V., Schramm, V.L., Zheng, W., and Wolberger, C. (2008). Structural insights into intermediate steps in the Sir2 deacetylation reaction. Structure 16, 1368–1377. Hol, W.G. (1985). Effects of the a-helix dipole upon the functioning and structure of proteins and peptides. Adv. Biophys. 19, 133–165. Holm, L., and Sander, C. (1993). Protein structure comparison by alignment of distance matrices. J. Mol. Biol. 233, 123–138. Hong, M., Fuangthong, M., Helmann, J.D., and Brennan, R.G. (2005). Structure of an OhrR-ohrA operator complex reveals the DNA binding mechanism of the MarR family. Mol. Cell 20, 131–141. Huffman, J.L., and Brennan, R.G. (2002). Prokaryotic transcription regulators: more than just the helix-turn-helix motif. Curr. Opin. Struct. Biol. 12, 98–106. Kim, J.H., Cho, E.J., Kim, S.T., and Youn, H.D. (2005). CtBP represses p300mediated transcriptional activation by direct association with its bromodomain. Nat. Struct. Mol. Biol. 12, 423–428. Kumar, P.R., Yu, Y., Sternglanz, R., Johnston, S.A., and Joshua-Tor, L. (2008). NADP regulates the yeast GAL induction system. Science 319, 1090–1092. Lamb, H.K., Stammers, D.K., and Hawkins, A.R. (2008). Dinucleotide-sensing proteins: linking signaling networks and regulating transcription. Sci. Signal. 1, pe38.

Molecular Cell Structural Mechanism of NADH/NAD+ Redox Sensing

Larsson, J.T., Rogstam, A., and von Wachenfeldt, C. (2005). Coordinated patterns of cytochrome bd and lactate dehydrogenase expression in Bacillus subtilis. Microbiology 151, 3323–3335.

Rutter, J., Reick, M., Wu, L.C., and McKnight, S.L. (2001). Regulation of clock and NPAS2 DNA binding by the redox state of NAD cofactors. Science 293, 510–514.

Laskowski, R.A., Moss, D.S., and Thornton, J.M. (1993). PROCHECK: a program to check the sterochemical quality of protein structures. J. Appl. Cryst. 26, 283–291.

Sanders, B.D., Zhao, K., Slama, J.T., and Marmorstein, R. (2007). Structural basis for nicotinamide inhibition and base exchange in Sir2 enzymes. Mol. Cell 25, 463–472.

Lewis, M. (2005). The lac repressor. C. R. Biol. 328, 521–548.

Schau, M., Chen, Y., and Hulett, F.M. (2004). Bacillus subtilis YdiH is a direct negative regulator of the cydABCD operon. J. Bacteriol. 186, 4585–4595.

Lim, D., Poole, K., and Strynadka, N.C. (2002). Crystal structure of the MexR repressor of the mexRAB-oprM multidrug efflux operon of Pseudomonas aeruginosa. J. Biol. Chem. 277, 29253–29259. Lin, S.J., Ford, E., Haigis, M., Liszt, G., and Guarente, L. (2004). Calorie restriction extends yeast life span by lowering the level of NADH. Genes Dev. 18, 12–16. Nakamura, A., Sosa, A., Komori, H., Kita, A., and Miki, K. (2007). Crystal structure of TTHA1657 (AT-rich DNA-binding protein; p25) from Thermus thermophilus HB8 at 2.16 A resolution. Proteins 66, 755–759. Nakahata, Y., Sahar, S., Astarita, G., Kaluzova, M., and Sassone-Corsi, P. (2009). Circadian control of the NAD+ salvage pathway by CLOCK-SIRT1. Science 324, 654–657. Nardini, M., Spano, S., Cericola, C., Pesce, A., Massaro, A., Millo, E., Luini, A., Corda, D., and Bolognesi, M. (2003). CtBP/BARS: a dual-function protein involved in transcription co-repression and Golgi membrane fission. EMBO J. 22, 3122–3130. Newberry, K.J., Fuangthong, M., Panmanee, W., Mongkolsuk, S., and Brennan, R.G. (2007). Structural mechanism of organic hydroperoxide induction of the transcription regulator OhrR. Mol. Cell 28, 652–664. Oppenhiemer, N.J., and Handlon, A.L. (1992). Mechanism of NAD-dependent enzymes. In The Enzymes: Mechanisms of Catalysis, D.S. Sigman, ed. (San Diego, CA: Academic Press, Inc), pp. 454–486. Pan, J.S., Hong, M.Z., and Ren, J.L. (2009). Reactive oxygen species: a double-edged sword in oncogenesis. World J. Gastroenterol. 15, 1702–1707. Pennella, M.A., and Giedroc, D.P. (2005). Structural determinants of metal selectivity in prokaryotic metal-responsive transcriptional regulators. Biometals 18, 413–428. Prasad, G.S., Wahlberg, M., Sridhar, V., Sundaresan, V., Yamaguchi, M., Hatefi, Y., and Stout, C.D. (2002). Crystal structures of transhydrogenase domain I with and without bound NADH. Biochemistry 41, 12745–12754. Rahangdale, S., Yeh, S.Y., Malhotra, A., and Veves, A. (2009). Therapeutic interventions and oxidative stress in diabetes. Front. Biosci. 14, 192–209. Ramsey, K.M., Yoshino, J., Brace, C.S., Abrassart, D., Kobayashi, Y., Marcheva, B., Hong, H.K., Chong, J.L., Buhr, E.D., Lee, C., et al. (2009). Circadian clock feedback cycle through NAMPT-mediated NAD+ biosynthesis. Science 324, 651–654. Reddy, V.P., Zhu, X., Perry, G., and Smith, M.A. (2009). Oxidative stress in diabetes and Alzheimer’s disease. J. Alzheimers Dis. 16, 763–774.

Seeman, N.C., Rosenberg, J.M., and Rich, A. (1976). Sequence-specific recognition of double helical nucleic acids by proteins. Proc. Natl. Acad. Sci. USA 73, 804–808. Semenza, G.L. (2009). Regulation of oxygen homeostasis by hypoxia-inducible factor 1. Physiology (Bethesda) 24, 97–106. Sickmier, E.A., Brekasis, D., Paranawithana, S., Bonanno, J.B., Paget, M.S., Burley, S.K., and Kielkopf, C.L. (2005). X-ray structure of a Rex-family repressor/NADH complex insights into the mechanism of redox sensing. Structure 13, 43–54. Sickmier, E.A., Frato, K.E., Shen, H., Paranawithana, S.R., Green, M.R., and Kielkopf, C.L. (2006). Structural basis for polypyrimidine tract recognition by the essential pre-mRNA splicing factor U2AF65. Mol. Cell 23, 49–59. Simon, M.C., Liu, L., Barnhart, B.C., and Young, R.M. (2008). Hypoxia-induced signaling in the cardiovascular system. Annu. Rev. Physiol. 70, 51–71. Thoden, J.B., Frey, P.A., and Holden, H.M. (1996). Crystal structures of the oxidized and reduced forms of UDP-galactose 4-epimerase isolated from Escherichia coli. Biochemistry 35, 2557–2566. Wang, E., Bauer, M.C., Rogstam, A., Linse, S., Logan, D.T., and von Wachenfeldt, C. (2008). Structure and functional properties of the Bacillus subtilis transcriptional repressor Rex. Mol. Microbiol. 69, 466–478. White, A., Ding, X., vanderSpek, J.C., Murphy, J.R., and Ringe, D. (1998). Structure of the metal-ion-activated diphtheria toxin repressor/tox operator complex. Nature 394, 502–506. Williamson, D.H., Lund, P., and Krebs, H.A. (1967). The redox state of free nicotinamide-adenine dinucleotide in the cytoplasm and mitochondria of rat liver. Biochem. J. 103, 514–527. Zhang, Q., Piston, D.W., and Goodman, R.H. (2002). Regulation of corepressor function by nuclear NADH. Science 295, 1895–1897. Zhao, Y., Zhang, J., Li, H., Li, Y., Ren, J., Luo, M., and Zheng, X. (2008). An NADPH sensor protein (HSCARG) down-regulates nitric oxide synthesis by association with argininosuccinate synthetase and is essential for epithelial cell viability. J. Biol. Chem. 283, 11004–11013. Zheng, X., Dai, X., Zhao, Y., Chen, Q., Lu, F., Yao, D., Yu, Q., Liu, X., Zhang, C., Gu, X., and Luo, M. (2007). Restructuring of the dinucleotide-binding fold in an NADP(H) sensor protein. Proc. Natl. Acad. Sci. USA 104, 8809–8814. Zhou, C., Huang, Y., and Przedborski, S. (2008). Oxidative stress in Parkinson’s disease: a mechanism of pathogenic and therapeutic significance. Ann. N Y Acad. Sci. 1147, 93–104.

Molecular Cell 38, 563–575, May 28, 2010 ª2010 Elsevier Inc. 575