A Eukaryotic-Like Interaction of Soluble Cyanobacterial Sensory Rhodopsin Transducer with DNA

doi:10.1016/j.jmb.2011.06.002

J. Mol. Biol. (2011) 411, 449–462 Contents lists available at www.sciencedirect.com

Journal of Molecular Biology j o u r n a l h o m e p a g e : h t t p : / / e e s . e l s e v i e r. c o m . j m b

A Eukaryotic-Like Interaction of Soluble Cyanobacterial Sensory Rhodopsin Transducer with DNA Shenlin Wang 1 , So Young Kim 2 , Kwang-Hwan Jung 2 , Vladimir Ladizhansky 1 and Leonid S. Brown 1 ⁎ 1 2

Department of Physics, University of Guelph, Ontario, Canada N1G 2W1 Department of Life Science and Institute of Biological Interfaces, Sogang University, Seoul 121-742, Korea

Received 13 April 2011; received in revised form 1 June 2011; accepted 2 June 2011 Available online 12 June 2011 Edited by M. F. Summers Keywords: photosensory transduction; transcription factors; NMR spectroscopy; DNA binding; secondary structure

Anabaena sensory rhodopsin is a recently discovered membrane photosensor with a unique signal transduction cascade. It interacts with a soluble tetrameric transducer [Anabaena sensory rhodopsin transducer (ASRT)] that can bind to promoter regions of several genes related to the utilization of light energy. Even though the X-ray crystal structure of ASRT is available, the mechanism of its interaction with DNA is still unknown. We used solution NMR to understand the mechanism of the DNA binding. Both Xray crystal structures and solution NMR data reveal seven β-strands forming a rigid scaffold (β-face) and a flexible, partially disordered α-face, comprised by the C-termini and loops. We found that the conformation of the α-face in solution is very different from that in the crystals. While the Ctermini of crystalline ASRT are solvent exposed and either α-helical or disordered, about half of ASRT monomers in solution feature buried Cterminal β-strand, with another half of C-tails being random coils. Titration of ASRT with a 20-bp fragment of the pec operon promoter showed that only monomers with β-structured C-tails bind the DNA. NMR signals suggest that specific Arg and Asn/Gln residues are involved in the interaction with DNA. The DNA binding occurs with micromolar affinity and a 1:1 stoichiometry (DNA:ASRT tetramer) and results in a significant ordering of the α-face involving the extension of the C-terminal β-strand and reorganization of the first loop. Such induced-fit type of interaction, which mainly utilizes loops between β-strands and results in the increase in their order, is typical for eukaryotic transcription factors of the immunoglobulin-like fold. © 2011 Elsevier Ltd. All rights reserved.

Introduction *Corresponding author. E-mail address: [email protected]. Abbreviations used: ASR, sensory rhodopsin from Anabaena (Nostoc) sp. PCC 7120; ASRT, Anabaena sensory rhodopsin transducer; CSI, chemical shift index; HSQC, heteronuclear single quantum coherence; NOESY, nuclear Overhauser enhancement spectroscopy; TROSY, transverse relaxation optimized spectroscopy; PDB, Protein Data Bank; NOE, nuclear Overhauser enhancement; dsDNA, double-stranded DNA.

Anabaena sensory rhodopsin (ASR) is the first photosensory rhodopsin discovered in eubacterial domain, in the genome of freshwater cyanobacterium Anabaena sp. PCC 7120.1 Similar to other rhodopsins, ASR is a membrane-embedded protein, composed of seven transmembrane helices, with covalently attached retinal chromophore.2,3 Unlike in its archaeal homologs, the cytoplasmic half of ASR is unusually hydrophilic and contains several water molecules forming a hydrogen-bonded network with polar side chains, which may be related

0022-2836/$ - see front matter © 2011 Elsevier Ltd. All rights reserved.

450 to its photosensory function.2,4,5 Depending on the light quality, ASR can change the isomeric state of retinal, switching between two stable conformations and displaying photochromism.2,6–8 Lightinduced structural changes in ASR9–11 may alter its interaction with the cognate transducer, starting the signal transduction cascade. It is believed that the ASR signal transduction cascade may regulate expression of various light-harvesting proteins and proteins related to the circadian clock, being responsible for the chromatic adaptation.12 A 14-kDa cytoplasmic soluble protein of 125 amino acids, Anabaena sensory rhodopsin transducer (ASRT), co-expresses with ASR in a single operon. 1 In solution, it exists in a compact tetrameric form.13 Biochemical in vitro data showed that ASRT tetramer binds ASR and affects its photochemistry.1,6,13 Thus, ASRT may function as the downstream partner of ASR in its lightactivated transduction cascade. It was suggested that photochemical transformations of ASR may result in the release of pre-bound ASRT, allowing it to interact with the downstream partners, such as DNA, by binding to the promoter regions of several photobiology-related genes and directly mediating gene expression.12 ASRT-like proteins are not unique to cyanobacteria. ASRT has many close homologs in various species of Bacteria and Archaea, which do not contain photoreceptors such as ASR.13,14 Instead, the genes encoding ASRT-like

Interaction of DNA and ASRT

proteins (so-called DUF1362 or ASRT family) are often found in clusters with those encoding proteins involved in sugar metabolism, as well as other membrane proteins related to bioenergetics. Structural bioinformatics analysis of ASRT homologs indeed suggests that they may bind small molecules, possibly sugars.14 Therefore, ASRT may be a representative of a new class of transducers that interact with DNA on one end and with membrane proteins and/or small metabolites on the other end. X-ray structures of four different crystal forms of the ASRT tetramer have been reported by Vogeley et al.13,15 They showed a predominance of β-strands, which form a rigid scaffold on one face of the transducer (the β-face), and flexible, partially disordered loops and C-tails on the other face (Fig. 1). This flexible α-face was suggested to directly interact with ASR.13 If ASRT interacts with ASR and DNA on the same face, its sequestration on the membrane can indeed regulate gene transcription. However, none of the ASRT structures revealed any obvious known DNA-binding motifs possibly because of the disorder of the DNA-binding regions in crystals. While most DNA-binding proteins have primarily α-helical fold, there are a number of protein transcription factors with predominant β-sheet architecture. Their interactions with DNA can be mediated either by β-strands themselves (e.g., TATA-binding proteins) or by flexible loops and termini

Fig. 1. Top: two views of the X-ray structure of the asymmetric ASRT tetramer (PDB code: 2II713) showing an overall architecture of the transducer. The two faces are indicated; individual monomers are shown in different colors. Bottom: secondary structure elements (β and L, β-strands and loops; η, a helical turn) are marked on the structure of ASRT monomer. The regions shown in purple are not detectable by NMR, probably due to their conformational plasticity.

451

Interaction of DNA and ASRT

associated with the strands (e.g., immunoglobulinlike fold proteins such as p53 and Runt domain of core binding factor).16–18 In the latter case, DNA binding often induces significant structural rearrangements, involving compactization and structuring of the disordered parts of the protein (the induced-fit mechanism).19,20 Thus, for definition of the exact mode of interaction of ASRT and DNA, structural comparison of the free and DNA-bound protein forms is required. In this work, we used solution NMR to characterize alternative conformations of noncrystalline ASRT and define its DNA-binding surface and structural changes induced by the binding. Multidimensional solution NMR has been very successful in providing structural information on interactions of DNA with protein transcription factors.21–25 We found that while the rigid βscaffold of ASRT tetramers looks similar in solution and in the crystal form, there is a large conformational variability in the flexible surface of the α-face. ASRT tetramer binds DNA on this flexible surface, as follows from the interaction study of ASRT with a 20-bp fragment of the pec operon promoter of Anabaena. The DNA binding leads to the increase in the structural order of the α-face of ASRT, suggesting the induced-fit mechanism. The binding mainly affects the loops between the β-strands, making it similar to that observed in eukaryotic transcription factors of the immunoglobulin-like fold.19,20,26–32

Results and Discussion Secondary structure of ASRT in solution X-ray structures of several crystal forms of ASRT have been reported by Vogeley et al.13 In crystals, ASRT is a tetramer with a doughnut-like shape. Each monomer consists of seven β-strands forming a rigid scaffold (so-called β-face) of a jelly roll β-sandwich fold (Fig. 1). On the other side, largely disordered loops and the C-termini constitute a flexible face called α-face, whose conformation depends on the crystallization conditions and on the position of the 6-His tag. Depending on a crystal form, the C-termini are either disordered (as evident from the missing electron density for residues R109–T124 comprising the C-tails) or seen as nonidentical solvent-exposed α-helices breaking the symmetry of the tetramer. Additionally, various parts of the first large loop (loop L-1 in Fig. 1) between the first and the second β-strands (β-1 and β-2, Fig. 1) are missing in most of the monomers (residues P18–E24 or Y19–E31). To compare the structure of ASRT in solution with that in crystals, we performed backbone resonance assignments and used chemical shift indexing (CSI)33 to assess the secondary structure of the protein. 1H–15N transverse relaxation optimized spectroscopy (TROSY) spectra of ASRT showed well-dispersed resonances, indicating a well-folded structure (Fig. 2). Backbone assignments were

Fig. 2. 1H–15N TROSY spectrum of free ASRT. The double resonances reflecting alternative conformations of the Cterminal part of the protein are colored green and red for the residues on the so-called F-tails and D-tails (see the text for details), respectively. Two resonances reflecting alternative conformations of G22 in the first loop are labeled as well.

452 obtained using standard strategies based on tripleresonance experiments (see Materials and Methods). The assignments of backbone amide resonances could be obtained for 88 out of the expected 113 nonprolyl residues, not counting the N-terminal 6-His tag. The majority of undetected residues (mapped on Fig. 1, bottom panel) are located on the flexible αface. The effective global correlation time, τc, estimated using the T1(15N)/T2(15N) ratio of the backbone amide groups of the residues located in the rigid β-scaffold of the protein, was found to be approximately 27 ns. This is comparable to that for globular proteins with a molecular mass of ∼ 60 kDa,34 confirming that ASRT is a tetramer in solution, as the molecular mass of monomeric isotope-labeled His-tagged ASRT is ∼ 14.7 kDa. This is consistent with the previous report that ASRT forms tetramers in solution, as measured by

Interaction of DNA and ASRT

size-exclusion chromatography13 and by our native SDS-PAGE data (not shown). A single set of resonances on the β-face was detected and assigned (71 out of the expected 81 residues), confirming that the four monomers of ASRT tetramer form a highly symmetric and rigid core. CSI33 identified seven β-strands and one helical turn for the residues located on the β-face (Fig. 3), mostly matching the secondary structure elements observed in the crystal structures of ASRT (Fig. 1).13 On the contrary, the α-face of ASRT, which is mainly composed of loop 1 and the C-termini, was found to be structurally disordered and flexible. Residues Q26–E31 of loop 1 and residues T104– A111 of the C-terminus (Fig. 1) could not be detected in the 1H–15N TROSY spectrum of ASRT, possibly due to the intermediate timescale conformational

Fig. 3. Analysis of the secondary structure of free and DNA-bound ASRTs using CSI.33 The secondary structure of the protein derived from X-ray is shown on top, and the amino acid sequence is shown at the bottom, with residues affected by the DNA binding highlighted yellow. The NMR data suggest the presence of two alternative conformations for the Cterminus: (a) random coil (F-tails) and (b) β-strand (D-tails, the DNA-binding conformation). DNA binding results in reorganization of the C-terminus of the D-conformers into a longer β-strand (shown in (c)) and in restructuring of the first loop. The latter can occur in either D- or F-conformers.

453

Interaction of DNA and ASRT

exchange. This observation is consistent with the lack of electron density for loop 1 and for the last 20 residues of the C-terminus in some of the crystal forms of ASRT.13 Furthermore, two sets of resonances with significantly different chemical shifts were detected for residues L116–T124 in the C-terminus and for residue G22 of loop 1 (Fig. 2), indicating the presence of two distinct conformations in these regions. Only one set of L116–T124 resonances could be linked to those of residues E112–L115. Thus, two different structures of the C-terminal part of ASRT could be deduced from the data. One conformation extends from residue E112 to T124 (called F-tail and F-monomer hereafter), and the other shorter structure includes residues L116 to T124 (D-tail and Dmonomer hereafter). Fewer resonances observed for the D-tails indicate the presence of additional intermediate timescale backbone conformational motions, which may cause severe line broadening for signals of some residues of the D-tails. Based on the CSI analysis, the F-tails are in the random-coil conformation (Fig. 3), while strongly negative CSI values obtained for residues L116 and S117 in the D-tails indicate the formation of either a β-strand or a similar H-bonded network involving the backbone in this region. Thus, the two alternative conformers differ dramatically in their C-terminal backbone secondary structures and, possibly, even in their overall tertiary structure (e.g., packing of the C-terminus against the β-scaffold). As shown in the 15 N-edited nuclear Overhauser enhancement spectroscopy (NOESY)–heteronuclear single quantum coherence (HSQC) spectra (Fig. 4a and b), residues from the D-tails have a significantly higher number of nuclear Overhauser enhancement (NOE) cross peaks than those from the F-tails, confirming that the D-monomers have a more ordered C-terminal structure than the F-monomers. Additionally, we observed a long-range NOE between backbone HN atoms of residues H103 and T118, located in β-strand 7 and D-tail, respectively, which indicates that the short β-strand 8 on D-tail may form a β-sheet with β-strand 7. Another longrange NOE cross peak was assigned to residues on D-tail and β-strand 1, that is, HN atoms of Y121 and A10 (Fig. 4c), showing the interactions of N-terminal and C-terminal parts of the protein. From the crystal structures of ASRT,13 β-strand 1 and β-strand 7 are located in the middle of the β-folded part of ASRT tetramer, spatially close to the central cavity. Thus, the proton proximities detected by the NOEs suggest that D-tails, parts of which organize into short β-strands, are buried inside the tetramer and traverse through the middle of the central cavity of ASRT. As a result, some C-terminal residues of Dtails should be very close (within 4–5 Å) to the Nterminus. Such D-tail conformation observed by solution NMR is very different from that detected in

the crystal structures, where the C-termini point away from the tetramer center cavity (Fig. 1). The distance between the end of the longest C-terminal α-helix observed in the crystals (backbone nitrogen of A120) and the backbone nitrogen of A10 is N35 Å, which is clearly inconsistent with the conformation suggested by NMR, where N- and C-tails are in close contact. Thus, while the β-face structure looks similar in crystals and in solution, the organization of the Ctermini is drastically different. In crystals, they are random coils or α-helices, depending on the crystallization conditions, and can vary between monomers in the same crystal form.13 In contrast, the C-terminal β-strand observed by NMR in the Dconformation of ASRT is a new structural feature, not seen in any of the crystal structures. This difference between the crystal and the solution states may arise from the different conditions of the samples and/or from intermolecular packing in crystals, which may force the flexible C-terminus to adopt an alternative conformation preferred under crystallization. Interestingly, the C-terminus of a homolog of ASRT, tetrameric TM1070 from Thermotoga maritima [Protein Data Bank (PDB) code: 1NC7], also adopts a β-stranded structure, similar to what we detected for ASRT. Moreover, the C-terminal βstrand of TM1070 forms antiparallel β-sheet with the first and penultimate β-strands, consistent with the structure of ASRT suggested by the long-range NOE pattern observed (Fig. 4c). Therefore, this short βstrand conformation of the C-terminus may be a common feature of ASRT-like proteins.13,14 Interaction of ASRT with DNA As shown above, in solution, ASRT is a tetramer, where each monomer can adopt two conformations: F-type with unstructured C-terminus and D-type with β-strand-bearing buried C-tail. We found that only C-terminal regions of D-monomers interact with DNA, while F-monomers remain unchanged. The interaction between ASRT and DNA was studied by titrating 15 N-labeled ASRT with an unlabeled 20-bp fragment of the pec operon, which includes five genes responsible for the biosynthesis of phycoerythrocyanin.35 The binding was followed by monitoring peaks' positions and intensities in the 1H–15N TROSY spectra. Additional assignment experiments were conducted on the ASRT:DNA complex, as described in Materials and Methods. The addition of DNA to the protein's solution induces further structuring of the C-terminal part of D-monomers. Seven additional residues, R109–L115 in the D-tails, could be detected and assigned (Fig. 5). The average intensity of the new resonances is about two times lower than that of the existing resonances in β-strands, in accordance with the idea

454 Interaction of DNA and ASRT

Fig. 4. 15N–1H strip plots from the three-dimensional 15N NOESY–HSQC spectra showing two alternative sets of resonances for the C-terminal part of ASRT. Strips of the 15N-edited NOESY–HSQC spectra of residues of D-tails (a) and F-tails (b). (c) Strips of 15N-edited NOESY–HSQC spectra containing NOEs between D-tail and βstrand 1 (Y121 and A10) and between D-tail and β-strand 7 (H103 and T118).

Interaction of DNA and ASRT

455

Fig. 5. 1H–15N TROSY–HSQC spectra of free (blue) and DNA-bound (red) ASRTs. The peaks, which appear or substantially change their positions upon the DNA binding, are marked. The assigned residues are found in the large loop between the first and the second β-strands and in the C-terminal part (only in the conformation with the terminal βstrand, so-called D-tails).

that only D-monomers interact with DNA. The CSI analysis shows that the D-tail undergoes rearrangement into a longer β-strand (Fig. 3c). Additional conformational changes are detected in loop 1, where residues F27–S29, undetectable in the free ASRT, could be identified in the DNA-bound conformation. Another residue, G22, exhibits two resonances (Fig. 5). The intensity of the G22 peak corresponding to the DNA-bound form builds up with increasing DNA concentration (Fig. 6), whereas the intensity of the cross peak due to free ASRT decreases. Both intensities become roughly equal when the DNA:tetramer ratio reaches unity. Finally, there is another DNA-induced peak appearing at 9.5/129.9 ppm (Figs. 5 and 6), which could not be assigned. Its high values of the 15N and 1H chemical shifts are consistent with the statistical trends observed for β-strands,36 suggesting that the corresponding residue may belong to a newly formed βstructure, possibly in the C-tail or in the first loop. We note that all of the assigned residues affected by the DNA binding are located on the exposed portion of the α-face of the tetramer, whereas the internal residues, including those directly involved in the inter-monomer interactions, remain unaffected. It is thus very likely that the DNA binding does not cause the dissociation of the tetramer. Estimation of τc for the DNA-bound ASRT complex shows its marginal decrease (27 ns versus 26 ns for free and DNA-bound ASRTs, respectively), which may be related to the induction of a more compact structure

of the α-face by the DNA, but is inconsistent with the tetramer dissociation. Such compactization of DNA-binding loops has been observed for several eukaryotic β-sandwich transcription factors of the

Fig. 6. Titration curves for selected resonances of ASRT upon addition of the 20-bp DNA fragment of the pec operon promoter. The normalized relative intensities of the 1 H–15N TROSY cross peaks at 298 K are plotted against the molar ratio of the added DNA and ASRT tetramer. Symbols: A114 is shown with stars, G22 is shown with circles, L115 is shown with hexagons, and an unassigned peak (at 9.5/129.9 ppm) is shown with triangles. The line represents a fit with the equation given in Materials and Methods, yielding a Kd of 1.7 ± 0.2 μM.

456 immunoglobulin-like fold, consistent with the induced-fit mechanism.19,20 The strength of the binding was estimated by following the intensity of newly appearing resonances upon titration of the ASRT solution with DNA (Fig. 6). The titration process was almost complete when the molar ratio between DNA and ASRT tetramer reached 1:1. Taken together with the lack of appreciable change in the correlation time (see above), this implies that each ASRT tetramer binds one DNA fragment, and the dissociation constant was estimated to be ∼ 1.7 ± 0.2 μM. While this apparent affinity is lower than that known for most DNA-binding proteins, it is in the same range as the affinity observed for transcription factors of similar β-sandwich architecture, such as the homodimer of nuclear factor κB p65 and monomeric nuclear factor of activated T-cells, cytoplasmic.19,37,38 The 1:1 binding stoichiometry and the fact that only the C-termini of D-monomers are affected by the DNA addition suggest that tetramers are composed of both D- and F-monomers and that the DNA interacts with the C-terminal parts of two out of four monomers. On the other hand, if one assumes that D- and F- monomers are confined to different tetramers, the observed relative intensities of the DNA-induced resonances (Fig. 5) and the corresponding titration curves (Fig. 6) would suggest that D-tetramers bind two DNA molecules, while F-tetramers bind none. This would result in two populations of complexes with different effec-

Interaction of DNA and ASRT

tive global correlation times, contrary to our experimental observations. Furthermore, considering that the intensity ratio between the G22 resonances of the free and DNA-bound ASRT forms is about unity, it is also feasible that only two of the four loops 1 in the tetramer are affected by the DNA. Thus, from the 1:1 binding stoichiometry and relative intensity of the new DNA-induced resonances, it follows that the DNA may interact with two loops 1 and two C-terminal D-tails in a tetramer, while the other two loops 1 and two F-tails remain in the unbound state. Protein–DNA interactions often occur through hydrogen bonding involving polar protein side chains.17,18,39 1 H–15 N HSQC spectra with large spectral window were collected to detect possible side-chain changes upon the DNA binding. As shown in Fig. 7, two additional pairs of side-chain resonances were detected when the DNA was added into the ASRT solution, with their 15 N chemical shifts at 73.8 and 99.8 ppm. The former value is typical for terminal nitrogens (Nη) of the guanidinium group of Arg, whereas the latter is typical for side-chain amide nitrogens of Gln or Asn residues. 40,41 One of the proton shifts of the guanidinium group is 9.2 ppm, indicating the formation of H-bond by this Arg side chain through one of the H η atoms, possibly with DNA. 42 Considering the amino acid sequence of the Cterminal region immobilized and structured by the DNA binding (Fig. 3), those side chains may

Fig. 7. 1H–15N HSQC spectra optimized for the detection of Arg, Gln, and Asn side-chain nitrogens. The spectra of DNA-free and DNA-bound forms are shown in blue and red, respectively. Proton resonances of the Arg and Gln/Asn nitrogens affected by the DNA binding are connected by horizontal lines.

Interaction of DNA and ASRT

represent R109 and Q110 (or N113) in the β-7/β8 linker of the D-tails. Indeed, those residues are spatially close to each other and may form the DNA

457 binding site (Fig. 8), comprised by the two side chains of the Arg residues (R109) of two different monomers and two amide groups of Gln and/or

Fig. 8. Structural model of ASRT monomer obtained from homology modeling on the template of the TM1070 homolog (PDB code: 1NC7) (a) in comparison with the X-ray-based model (PDB code: 2II7) (b). Protons of the residues for which longrange NOEs were observed (Fig. 4c) are shown as purple balls, with the distances measured from the models indicated. The region of contacts of the C-terminal β-strand with the first and seventh β-strands (marked) is enlarged. The residues affected by the DNA are colored yellow. (c) ASRT tetramer model produced by replacing two monomers from the crystal structure of ASRT (PDB code: 2II7) with two monomers obtained from modeling on the template of the TM1070 homolog (PDB code: 1NC7), in the opposing positions (an alternative model with adjacent positions is possible). Monomers from the crystal structure (representing F-monomers) are shown in gray, while modeled monomers (representing D-monomers) are shown in cyan. Side chains of putative DNA-binding residues are shown: Arg109 side chains are shown in red, and Gln110 and Asn113 side chains are in green. The backbone of residues affected by the DNA binding is shown in yellow.

458 Asn residues, that is, Q110 and/or N113, from the same monomers. It is not clear if any residues from loop 1 are involved in direct binding of DNA or it is simply restructured as a result of changed interactions with the elongated C-terminal β-strand or steric clash with the DNA. To visualize the putative binding site for DNA, we mapped the residues affected by the DNA binding on the three-dimensional model of ASRT solution structure. As the X-ray structures of ASRT either do not show C-terminal parts or have them dramatically different (i.e., helical) from those detected by NMR, a different structure had to be used as a basis for modeling. Similar to what was done previously,13 the ASRT structure was modeled on the template of its homolog from T. maritima TM1070 (PDB code: 1NC7), which has similar C-terminal β-strands. Structural model of ASRT monomer obtained by this procedure (Fig. 8a) showed that residues A114–I119 form an additional β-strand (β-8), consistent with the observed CSI for the residues of D-tails (Fig. 3). Two alternative ASRT tetramer models were built by replacing two of the four monomers in the X-ray crystal structure of ASRT (PDB code: 2II7), representing F-monomers, with two monomers modeled on the 1NC7 template, representing D-monomers in either opposing or adjacent positions (Fig. 8c; only one of the models is shown). As can be seen in the models, the β-8 strands of D-monomers are largely buried inside the tetramer core (Fig. 8c) and form antiparallel β-sheets with β-7 and β-1 strands (Fig. 8a), different from what was detected in the crystal structures,13 where the observed C-termini are solvent exposed (Fig. 1). The estimated distances between HN atoms of Y121 and A10 and between HN atoms of H103 and T118 are 4 Å and 2.8 Å (Fig. 8a), respectively, matching the observations from the NOESY spectrum (Fig. 4c), while the corresponding distances in the X-ray-based model are N 35 and N 27 Å (Fig. 8b). The disordered F-tails of the other two monomers could be solvent exposed, as observed in the crystal structure and confirmed by the lack of the long-range NOEs with the buried β-strands. The residues affected by the DNA binding are located on the flexible α-face, and some of their polar side chains are solvent exposed, potentially forming the DNA binding site. The residues from the D-tails strongly affected by the DNA binding comprise the linker between the β-7 and the β-8 strands (Fig. 8). Thus, the β-stranded structure of the D-tails may be an important factor defining the correct conformation of these linkers, allowing them to recognize their DNA partners. On the contrary, homologous regions of the F-tails are structurally disordered, making them unfit for the DNA binding. The reorganization of the D-tails upon the DNA binding must affect loops 1 of the same monomers (the

Interaction of DNA and ASRT

distance is 9 Å between loop 1 and the β-7/β8 linker, measured between backbone nitrogen atoms of residues R109 and F27), restricting their conformational mobility and making them visible in the NMR spectra. The resonances for a number of residues remain missing from the spectra even upon association of ASRT with the DNA, that is, those for residues G14– I16 and H30–T32 in loop 1 and residues T104–N107 in the C-terminal part, indicating that the backbone conformational plasticity for those residues is still present. From the crystal structures, those residues are located either at the edge of loop 1 or in the linker connecting the β-face and the C-terminus. Therefore, even though the DNA binds with ASRT strongly, the interdomain motions between the DNA-binding regions and the core of ASRT tetramer are very likely. Interestingly, the DNA fragment used in this study is asymmetric. Asymmetric DNA targets are known to break the symmetry of DNA-binding proteins, such as HAP-1 homodimers and related proteins.43 On the other hand, some symmetric tetrameric gene regulators, such as TtgV,44 can bind DNA in a highly asymmetric fashion. As ASRT tetramers include two different conformations of monomers along with substantial flexible regions, it may be designed to interact with asymmetric DNA sequences efficiently. An inherent asymmetry of ASRT tetramer is further enhanced by DNA binding, which orders D-tails but not F-tails. The observed mode of interaction of ASRT with DNA is similar to the one known for eukaryotic βsandwich transcription factors of immunoglobulinlike fold. This family of DNA-binding proteins includes important developmental, immunological, and anticancer regulators such as p53, NFATC1, nuclear factor κB p50, STAT-1, and core binding factor Runt domain.19,20,26–32 Their DNA binding is characterized by participation of unstructured loops between β-strands, which become structured and compact as a result of the interaction with DNA (the induced-fit mechanism).19,20 Binding of DNA by loops of the β-sandwich of ASRT and structuring of its β1/β2 and β7/β8 linkers observed upon interaction with DNA strongly resemble the eukaryotic DNA binding. An additional similarity between DNA binding by ASRT and these eukaryotic transcription factors is that it is mainly mediated by positively charged (Arg and Lys) and hydrogen-bonding (Asn and Gln) residues. Finally, both ASRT and many of the eukaryotic transcription factors interact with DNA as oligomers, where some subunits (often two) are involved in direct interaction with DNA, while other subunits (or domains) ensure proper oligomerization, leading to the unique DNA-binding geometry.20,30–32

459

Interaction of DNA and ASRT

Summary and Conclusions ASRT is a recently discovered transducer of Anabaena sensory rhodopsin, possibly serving as an intermediate component in a light-activated gene regulation system. Upon light absorption, membrane-embedded photochromic ASR alters its conformation and may change its affinity for ASRT, modulating its interactions with other biomacromolecules, such as DNA, and regulating cellular processes as a result. Using solution NMR, we found that ASRT tetramers in solution have a rigid scaffold composed of β-sheets of the monomers and a flexible α-surface, similar to those found by X-ray crystallography.13 Backbone conformational variations, including multiple conformation exchange for loop 1 and a part of the C-terminus, and structural heterogeneity of the Ctermini were characterized. Different from what was observed in the crystal structures, buried terminal βstrands were found for the half of the C-tails (so-called D-tails), while either random-coil or nonidentical helices were detected by X-ray. The DNA-binding study showed that ASRT strongly interacts with a 20-bp fragment of the pec operon promoter through its flexible surface, which becomes more ordered upon the DNA binding. Only two out of the four monomers of the ASRT tetramer, those with the β-structured tails, interact with the DNA. The interaction involves two C-terminal regions, whose βstrands become longer, and two loops between the first and the second β-strands. Hydrogen-bonding interactions of DNA with several polar side chains seem to be involved in the complex formation, possibly including the guanidinium groups of R109 and the amide groups of Q110 or N113 of opposing monomers in the D-tails' conformation. Although some local backbone motions of ASRT are frozen by the DNA binding, allowing the detection of resonances for a number of residues that are missing in the spectra of DNA-free ASRT, the slow motions are still present, as about a half of the originally undetected residues remain missing. Since the DNA binds to these flexible loops and tails, it is expected that the bound DNA may move relatively freely with respect to the core of ASRT. Binding of DNA on the flexible αface supports the idea that sequestration of ASRT by ASR may regulate gene expression in Anabaena. The induced-fit mechanism of interaction, where loops between β-strands become ordered upon DNA binding, is typical for eukaryotic transcription factors of the immunoglobulin-like fold.19,20

Materials and Methods Protein sample preparation ASRT was overexpressed in Escherichia coli strain BL21Codon Plus-RIL harboring plasmid containing the ASRT

coding sequence with an additional 6-His tag at the Nterminus, under IPTG-inducible promoter. The cells were grown in minimal medium containing 15NH4Cl and/or 13 C-labeled glucose for the production of labeled samples. Protein purification was performed by following the batch protocol of the Qiagen Ni2+-NTA manual. The final yields varied from 15 mg to 20 mg per liter of culture. For producing fractionally randomly deuterated samples, cells were grown in 50% D2O-based medium, with protein yield similar to that in H2O-based medium. The sample purity was assessed by SDS-PAGE and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. The purified ASRT samples were dialyzed against a buffer containing 50 mM Tris, 50 mM NaCl, and 10% D2O at pH 7.3 and concentrated to ∼ 1 mM (monomer concentration) for the NMR study. Protein concentrations were determined by measuring OD280 nm using an extinction coefficient of 13,610 M− 1 cm− 1.13 Preparation of double-stranded DNA for interaction studies As ASR was implicated in regulation of chromatic adaptation,1,12 promoters of a number of candidate genes (including pec, cpc, and kai ABC) were tested for interaction with ASRT using gel-shift and chromatin immunoprecipitation assays and further confirmed by isothermal titration calorimetry (S.Y.K., S. R. Yoon, S. I. Han, Y. Yoon, and K.-H.J., unpublished results). Further, the pec operon promoter region (80 bp) was broken into four 20-bp fragments, and the fragment with the highest affinity for ASRT was chosen. The corresponding oligonucleotide 5′-ATGACCTTTAGGAGGAAAGA-3′ and its complementary single-stranded DNA were purchased from Sigma-Genosys company. The synthesized singlestranded DNA oligonucleotides were dissolved in autoclaved milli-Q water and mixed stoichiometrically. The mixture was heated to 94 °C and gradually cooled down to 55 °C in 1 h. The prepared double-stranded DNA (dsDNA) oligonucleotides were lyophilized, and the powder was dissolved in the same buffer as that used for ASRT. The concentrations of dsDNA were determined by measuring OD260 nm using an extinction coefficient of 0.02 μg− 1 ml cm− 1. NMR spectroscopy All NMR data were collected on a Bruker Avance II spectrometer operating at a proton frequency of 600.130 MHz and equipped with a cryoprobe. Partially deuterated (∼50%), U-13C/15N-labeled sample was used for assignments. A set of TROSY-based triple-resonance experiments [HNCA, HN(CO)CA, HNCACB, HN(CO) CACB, HNCO, and HN(CA)CO]45,46 was performed for obtaining backbone assignments of free and DNA-bound ASRTs. All NMR experiments were done at 37 °C. 15Nedited NOESY–HSQC47,48 spectra were collected on free ASRT with a 200-ms mixing time. NMR samples showed no signs of degradation during the course of these experiments. Carbon chemical shifts were corrected for deuterium isotope effects prior to CSI calculation.49 15 N relaxation parameters were measured on 0.4 mM free and DNA-bound ASRTs by using TROSY-based pulse sequences at 25 °C.50 The relaxation times, T1 and T2, were

460

Interaction of DNA and ASRT

determined by fitting the cross-peak heights, obtained through the standard routine of the SPARKY program,51 measured as a function of the delay within the pulse sequence to a single-exponential decay. The effective τc value was estimated using the equations from the HYDRONMR program.52 Protein–DNA interactions For dsDNA titration experiments, a 15N-labeled ASRT sample was prepared at a concentration of 0.2 mM (monomer). Concentrated DNA solution was added stepwise to ASRT sample to reach a dsDNA:ASRT tetramer molar ratio of 0.25, 0.5, 0.75, 1.0, 2.0, and 4.0. The changes in the conformation of ASRT upon the DNA binding were followed using 1H–15N TROSY spectra at 25 °C. Selected cross peaks corresponding to residues newly detected upon DNA binding, including G22, A114, L115, and an unassigned peak (at 9.5/129.9 ppm), were integrated, with the intensities corrected for dilution and normalized using the intensities obtained at a dsDNA:ASRT tetramer molar ratio of 4:1. The dissociation constant (Kd) was obtained by fitting the normalized relative intensities against the molar ratio between dsDNA and ASRT tetramer using the equation: y=

ðcASRT + x
cASRT + Kd Þ −

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðcASRT + x
cASRT + Kd Þ2 − 4x
c2ASRT 2cASRT
P

where y is the normalized relative intensity, x is the molar ratio of dsDNA and ASRT tetramer, cASRT is the total concentration of ASRT tetramer, and P is the constant describing the normalized relative intensity when the molar ratio between dsDNA and ASRT is infinite. Structural modeling Structural model of monomeric ASRT was calculated using an online homolog modeling program, 3DJIGSAW,53 using the crystal structure of TM1070 of T. maritima (PDB code: 1NC7; chain A) as a template. ASRT tetramer models were made by replacing two monomers in the crystal structure of ASRT (PDB code: 2II7) with the modeled monomers in either opposing or adjacent positions. NMR assignment data The NMR assignment data are deposited in the BioMagResBank†, under accession numbers 17590, 17951, and 17952.

Acknowledgements The work was supported by the Natural Science and Engineering Research Council of Canada, the † http://www.bmrb.wisc.edu

Canada Foundation for Innovation, and the Ontario Ministry of Research and Innovation grants to V.L. and L.S.B and Korea Research Fund grants (3312008-1-C00242 and 2010-616-C00029) to K.-H.J. V.L. holds Canada Research Chair Tier II in Biophysics. S.W. was supported by the Canadian Institutes of Health Research fellowship.

References 1. Jung, K. H., Trivedi, V. D. & Spudich, J. L. (2003). Demonstration of a sensory rhodopsin in eubacteria. Mol. Microbiol. 47, 1513–1522. 2. Vogeley, L., Sineshchekov, O. A., Trivedi, V. D., Sasaki, J., Spudich, J. L. & Luecke, H. (2004). Anabaena sensory rhodopsin: a photochromic color sensor at 2.0 Å. Science, 306, 1390–1393. 3. Shi, L., Kawamura, I., Jung, K. H., Brown, L. S. & Ladizhansky, V. (2011). Conformation of a sevenhelical transmembrane photosensor in the lipid environment. Angew. Chem., Int. Ed. Engl. 50, 1302–1305. 4. Kawanabe, A. & Kandori, H. (2009). Photoreactions and structural changes of Anabaena sensory rhodopsin. Sensors, 9, 9741–9804. 5. Kouyama, T. & Murakami, M. (2010). Structural divergence and functional versatility of the rhodopsin superfamily. Photochem. Photobiol. Sci. 9, 1458–1465. 6. Sineshchekov, O. A., Trivedi, V. D., Sasaki, J. & Spudich, J. L. (2005). Photochromicity of Anabaena sensory rhodopsin, an atypical microbial receptor with a cis-retinal light-adapted form. J. Biol. Chem. 280, 14663–14668. 7. Kawanabe, A., Furutani, Y., Jung, K. H. & Kandori, H. (2007). Photochromism of Anabaena sensory rhodopsin. J. Am. Chem. Soc. 129, 8644–8649. 8. Strambi, A., Durbeej, B., Ferre, N. & Olivucci, M. (2010). Anabaena sensory rhodopsin is a light-driven unidirectional rotor. Proc. Natl Acad. Sci. USA, 107, 21322–21326. 9. Bergo, V. B., Ntefidou, M., Trivedi, V. D., Amsden, J. J., Kralj, J. M., Rothschild, K. J. & Spudich, J. L. (2006). Conformational changes in the photocycle of Anabaena sensory rhodopsin. Absence of the Schiff base counterion protonation signal. J. Biol. Chem. 281, 15208–15214. 10. Shi, L. C., Yoon, S. R., Bezerra, A. G., Jung, K. H. & Brown, L. S. (2006). Cytoplasmic shuttling of protons in Anabaena sensory rhodopsin: implications for signaling mechanism. J. Mol. Biol. 358, 686–700. 11. Kawanabe, A., Furutani, Y., Yoon, S. R., Jung, K. H. & Kandori, H. (2008). FTIR study of the L intermediate of Anabaena sensory rhodopsin: structural changes in the cytoplasmic region. Biochemistry, 47, 10033–10040. 12. Jung, K. H. (2007). The distinct signaling mechanisms of microbial sensory rhodopsins in Archaea, Eubacteria and Eukarya. Photochem. Photobiol. 83, 63–69. 13. Vogeley, L., Trivedi, V. D., Sineshchekov, O. A., Spudich, E. N., Spudich, J. L. & Luecke, H. (2007). Crystal structure of the Anabaena sensory rhodopsin transducer. J. Mol. Biol. 367, 741–751.

461

Interaction of DNA and ASRT 14. De Souza, R. F., Iyer, L. M. & Aravind, L. (2009). The Anabaena sensory rhodopsin transducer defines a novel superfamily of prokaryotic small-molecule binding domains. Biol. Direct. 4, 25; discussion 25. 15. Vogeley, L. & Luecke, H. (2006). Crystallization, X-ray diffraction analysis and SIRAS/molecular-replacenent phasing of three crystal forms of Anabaena sensory rhodopsin transducer. Acta Crystallogr., Sect. F: Struct. Biol. Cryst. Commun. 62, 388–391. 16. Luscombe, N. M., Austin, S. E., Berman, H. M. & Thornton, J. M. (2000). An overview of the structures of protein–DNA complexes. Genome. Biol. 1; REVIEWS001. 17. Shrivastava, T. & Tahirov, T. H. (2010). Threedimensional structures of DNA-bound transcriptional regulators. Methods Mol. Biol. 674, 43–55. 18. Garvie, C. W. & Wolberger, C. (2001). Recognition of specific DNA sequences. Mol. Cell, 8, 937–946. 19. Zhou, P., Sun, L. J., Dotsch, V., Wagner, G. & Verdine, G. L. (1998). Solution structure of the core NFATC1/ DNA complex. Cell, 92, 687–696. 20. Joerger, A. C. & Fersht, A. R. (2008). Structural biology of the tumor suppressor p53. Annu. Rev. Biochem. 77, 557–582. 21. Clore, G. M. & Gronenborn, A. M. (1994). Structures of larger proteins, protein ligand and protein–DNA complexes by multidimensional heteronuclear NMR. Protein Sci. 3, 372–390. 22. Otting, G., Qian, Y. Q., Billeter, M., Muller, M., Affolter, M., Gehring, W. J. & Wuthrich, K. (1990). Protein–DNA contacts in the structure of a homeodomain–DNA complex determined by nuclear magnetic resonance spectroscopy in solution. EMBO J. 9, 3085–3092. 23. Kalodimos, C. G., Boelens, R. & Kaptein, R. (2004). Toward an integrated model of protein–DNA recognition as inferred from NMR studies on the Lac repressor system. Chem. Rev. 104, 3567–3586. 24. Hard, T. (1999). NMR studies of protein–nucleic acid complexes: structures, solvation, dynamics and coupled protein folding. Q. Rev. Biophys. 32, 57–98. 25. Bagby, S., Arrowsmith, C. H. & Ikura, M. (1998). New perceptions of transcription factor properties from NMR. Biochem. Cell Biol. 76, 368–378. 26. Perez-Alvarado, G. C., Munnerlyn, A., Dyson, H. J., Grosschedl, R. & Wright, P. E. (2000). Identification of the regions involved in DNA binding by the mouse PEBP2 alpha protein. FEBS Lett. 470, 125–130. 27. Berardi, M. J., Sun, C., Zehr, M., Abildgaard, F., Peng, J., Speck, N. A. & Bushweller, J. H. (1999). The Ig fold of the core binding factor alpha Runt domain is a member of a family of structurally and functionally related Ig-fold DNA-binding domains. Structure, 7, 1247–1256. 28. Tahirov, T. H., Inoue-Bungo, T., Morii, H., Fujikawa, A., Sasaki, M., Kimura, K. et al. (2001). Structural analyses of DNA recognition by the AML1/Runx-1 Runt domain and its allosteric control by CBFbeta. Cell, 104, 755–767. 29. Nagata, T., Gupta, V., Sorce, D., Kim, W. Y., Sali, A., Chait, B. T. et al. (1999). Immunoglobulin motif DNA recognition and heterodimerization of the PEBP2/ CBF Runt domain. Nat. Struct. Biol. 6, 615–619. 30. Chen, X., Vinkemeier, U., Zhao, Y., Jeruzalmi, D., Darnell, J. E., Jr & Kuriyan, J. (1998). Crystal structure

31. 32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

of a tyrosine phosphorylated STAT-1 dimer bound to DNA. Cell, 93, 827–839. Ghosh, G., van Duyne, G., Ghosh, S. & Sigler, P. B. (1995). Structure of NF-kappa B p50 homodimer bound to a kappa B site. Nature, 373, 303–310. Muller, C. W., Rey, F. A., Sodeoka, M., Verdine, G. L. & Harrison, S. C. (1995). Structure of the NF-kappa B p50 homodimer bound to DNA. Nature, 373, 311–317. Wishart, D. S. & Sykes, B. D. (1994). The 13C chemicalshift index: a simple method for the identification of protein secondary structure using 13C chemical-shift data. J. Biomol. NMR, 4, 171–180. Tian, C., Vanoye, C. G., Kang, C., Welch, R. C., Kim, H. J., George, A. L., Jr. & Sanders, C. R. (2007). Preparation, functional characterization, and NMR studies of human KCNE1, a voltage-gated potassium channel accessory subunit associated with deafness and long QT syndrome. Biochemistry, 46, 11459–11472. Jung, L. J., Chan, C. F. & Glazer, A. N. (1995). Candidate genes for the phycoerythrocyanin alpha subunit lyase. Biochemical analysis of pecE and pecF interposon mutants. J. Biol. Chem. 270, 12877–12884. Wang, Y. & Jardetzky, O. (2002). Probability-based protein secondary structure identification using combined NMR chemical-shift data. Protein Sci. 11, 852–861. de Lumley, M., Hart, D. J., Cooper, M. A., Symeonides, S. & Blackburn, J. M. (2004). A biophysical characterisation of factors controlling dimerisation and selectivity in the NF-kappaB and NFAT families. J. Mol. Biol. 339, 1059–1075. Phelps, C. B., Sengchanthalangsy, L. L., Malek, S. & Ghosh, G. (2000). Mechanism of kappa B DNA binding by Rel/NF-kappa B dimers. J. Biol. Chem. 275, 24392–24399. Luscombe, N. M., Laskowski, R. A. & Thornton, J. M. (2001). Amino acid-base interactions: a three-dimensional analysis of protein–DNA interactions at an atomic level. Nucleic Acids Res. 29, 2860–2874. Pellecchia, M., Wider, G., Iwai, H. & Wuthrich, K. (1997). Arginine side chain assignments in uniformly 15 N-labeled proteins using the novel 2D HE(NE) HGHH experiment. J. Biomol. NMR, 10, 193–197. Liu, A. Z., Li, Y., Yao, L. S. & Yan, H. G. (2006). Simultaneous NMR assignment of backbone and side chain amides in large proteins with IS-TROSY. J. Biomol. NMR, 36, 205–214. Liu, A., Hu, W., Majumdar, A., Rosen, M. K. & Patel, D. J. (2000). NMR detection of side chain-side chain hydrogen bonding interactions in 13C/15N-labeled proteins. J. Biomol. NMR, 17, 305–310. King, D. A., Zhang, L., Guarente, L. & Marmorstein, R. (1999). Structure of a HAP1–DNA complex reveals dramatically asymmetric DNA binding by a homodimeric protein. Nat. Struct. Biol. 6, 64–71. Lu, D., Fillet, S., Meng, C., Alguel, Y., Kloppsteck, P., Bergeron, J. et al. (2010). Crystal structure of TtgV in complex with its DNA operator reveals a general model for cooperative DNA binding of tetrameric gene regulators. Genes Dev. 24, 2556–2565. Salzmann, M., Wider, G., Pervushin, K., Senn, H. & Wuthrich, K. (1999). TROSY-type triple-resonance

462 experiments for sequential NMR assignments of large proteins. J. Am. Chem. Soc. 121, 844–848. 46. Pervushin, K., Riek, R., Wider, G. & Wuthrich, K. (1997). Attenuated T2 relaxation by mutual cancellation of dipole–dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc. Natl Acad. Sci. USA, 94, 12366–12371. 47. Davis, A. L., Keeler, J., Laue, E. D. & Moskau, D. (1992). Experiments for recording pure-absorption heteronuclear correlation spectra using pulsed field gradients. J. Magn. Reson. 98, 207–216. 48. Nietlispach, D., Clowes, R. T., Broadhurst, R. W., Ito, Y., Keeler, J., Kelly, M. et al. (1996). An approach to the structure determination of larger proteins using triple resonance NMR experiments in conjunction with random fractional deuteration. J. Am. Chem. Soc. 118, 407–415.

Interaction of DNA and ASRT 49. Gardner, K. H., Rosen, M. K. & Kay, L. E. (1997). Global folds of highly deuterated, methyl-protonated proteins by multidimensional NMR. Biochemistry, 36, 1389–1401. 50. Zhu, G., Xia, Y. L., Nicholson, L. K. & Sze, K. H. (2000). Protein dynamics measurements by TROSY-based NMR experiments. J. Magn. Reson. 143, 423–426. 51. Goddard, T. D. & Kneller, D. G. SPARKY 3. University of California, San Francisco, CA. 52. de la Torre, J. G., Huertas, M. L. & Carrasco, B. (2000). HYDRONMR: prediction of NMR relaxation of globular proteins from atomic-level structures and hydrodynamic calculations. J. Magn. Reson. 147, 138–146. 53. Bates, P. A., Kelley, L. A., MacCallum, R. M. & Sternberg, M. J. (2001). Enhancement of protein modeling by human intervention in applying the automatic programs 3D-JIGSAW and 3D-PSSM. Proteins, 5, 39–46.