The Structure of the Coliphage HK022 Nun Protein-λ-phageboxBRNA Complex

Nun protein from coliphage HK022 binds to phageboxB RNA and functions, in contrast to phage λ N protein, as a transcriptional terminator. The basic Nun-(10–44) peptide contains the boxB RNA binding arginine rich motif, ARM. The peptide binds boxB RNA and competes with the phage λ ARM peptide N-(1–36) as indicated by nuclear magnetic resonance (NMR) spectroscopy titrations. In two-dimensional nuclear Overhauser enhancement spectroscopy experiments boxB RNA in complex with Nun-(20–44) exhibits the same pattern of resonances as it does in complex with N peptides containing the ARM, and we could show that Nun-(20–44) forms a bent α-helix upon binding to theboxB RNA. The structure of the boxB RNA-bound Nun-(20–44) was determined on the basis of 191 intra- and 30 intermolecular distance restraints. Ser-24 is anchored to the lower RNA stem, and stacking of Tyr-39 and A7 is clearly experimentally indicated. Arg-28 shows numerous contacts to the RNA stem. Leu-22, Ile-30, Trp-33, Ile-37, and Leu-41 form a hydrophobic surface, which could be a recognition site for additional host factors such as NusG. Such a hydrophobic surface area is not present in N-(1–36) bound toboxB RNA.

Bacteriophage N protein plays an essential role in transcriptional antitermination in the two-phage early operons that are critical for phage development. The inhibition of termination at intrinsic and -dependent terminators by N protein depends on recognition of an RNA element called nut 1 (N utilization) on the nascent phage transcript and on four Escherichia coli host factors (NusA, NusB, NusG, and ribosomal protein NusE). Together they form a ribonucleoprotein complex that converts the RNA polymerase into a termination-resistant form upon binding (1,2). N protein binds specifically with high affinity to phage boxB RNA, a 15-mer RNA hairpin containing a purine-rich pentaloop (3,4).
Nun protein of phage HK022 is a transcription termination factor that acts, in contrast to other termination factors, highly template-and site-specific. Nun terminates transcription uniquely on phage templates (5), competing with N protein for a common binding site, nut boxB RNA (6). Like N protein, Nun requires additional host factors (NusA, NusB, NusE, and NusG) for efficient termination, whereas the presence of NusA alone inhibits the termination activity of Nun (7). Recently it has been proposed that Nun arrests transcription by anchoring RNA polymerase to DNA (8). Both Nun and N proteins belong to the family of arginine-rich motif (ARM) binding proteins. The structures of phage N ARM peptide-boxB RNA complexes and of a phage P22 ARM N peptide-boxB RNA complex have been solved by NMR (9 -12). For both these phage peptides, a very similar mode of binding has been observed, with the peptides bound in the major groove of boxB RNA, which adopts a typical hairpin conformation closed by an apical tetraloop.
NMR Spectroscopy-All NMR experiments were recorded at 28°C for the Nun-(10 -44)-boxB RNA complex and at 30°C for the Nun-(20 -44)-boxB RNA complex on a Bruker DRX 600 spectrometer equipped with 1 H/ 13 C/ 15 N probes and triple-axis pulsed field gradient capabilities. For resonance assignment correlated spectroscopy (COSY), total coherence spectroscopy (TOCSY), and nuclear Overhauser enhancement spectroscopy (NOESY) experiments were performed using standard techniques for recording and water suppression (13). For the Nun-(10 -44)-boxB RNA complex, TOCSY experiments were recorded with 40 and 80 ms mixing time, and NOESY experiments were recorded with 80 and 200 ms mixing time, respectively. For the Nun-(20 -44)-boxB RNA complex, TOCSY experiments were recorded with a mixing time of 80 ms, and NOESY experiments were recorded with mixing time 150 and 300 ms. 1 H 15 N HSQC spectra were recorded with the fast HSQC pulse scheme (14). All NMR data were analyzed with the NDee (SpinUp Inc., Dortmund, Germany) and XWINNMR (Bruker, Karlsruhe, Germany) program packages augmented with in-house-written routines. Proton chemical shifts were referenced to external 2,2-dimethyl-2-silapentanesulfonic acid. The chemical shifts of the 15 N resonances were referenced indirectly using the 15 N/ 1 H ⌶ ratio of 0.10132905 of the zero-point frequency at 298 K (15).
Interproton distance restraints were obtained from two-dimensional NOESY spectra of both Nun peptide-boxB RNA complexes. NOE intensities were estimated semi-quantitatively on the basis of cross-peak intensities from NOESY spectra collected with 80 ms mixing time. The categories "strong", "medium", and "weak" were converted into distance constraints with upper limit of 2.7, 3.5, and 5.0 Å for peptide intramo-lecular NOEs (16). For peptide-RNA intermolecular NOEs, these categories were converted in distance constraints with upper limit of 3.0, 4.0, and 5.0 Å, respectively. NOEs that were only visible in the NOESY spectra of Nun-(10 -44)-boxB RNA complex with 200 ms mixing time were classified as "very weak" with an upper bound of 6.0 Å (10). To improve the convergence of the structure calculation, the lower bounds of all NOE restraints were set to 0 Å (17).
Molecular Dynamics Calculations-Experimental data clearly indicate that the boxB RNA in the Nun-(10 -44) complex is virtually identical to the boxB RNA in the N-(1-36) complex. We thus used the structure of the boxB RNA in the N-(1-36)-boxB RNA complex that we determined earlier (12) (Protein Data Bank entry 1qfq) as a fixed template for all molecular dynamics calculations.
All structure calculations were performed using a modified ab initio simulated annealing protocol with an extended version of X-PLOR 3.851 (18). The calculation strategy, which was described in detail previously (19), included floating assignment of prochiral groups (20), a conformational data base potential term (21), and a reduced presentation for non-bonded interactions for part of the calculation (19).
The conformational search phase (60 ps of molecular dynamics at 2000 K) was followed by cooling from 2000 K to 1000 K within 40 ps, concomitantly increasing the force constants for the non-bonded interactions and the angle energy constant for the diastereospecifically unassigned groups to their final values. In the next stage of the calculation, the system was cooled from 1000 K to 100 K within 30 ps, applying the high force constants obtained at the end of the previous cooling stage. To detect the energy minimum, 1200 steps of energy minimization were performed, the final 1000 steps without conformational data base potential. In the final round, 100 structures were calculated, and the 20 structures that showed the lowest energy and the least number of violations of the experimental data were selected for further characterization.
For geometrical analysis and investigation of secondary structure and structural parameters the PROCHECK (22) and NUCPLOT (23) programs were used. Quick visualization and graphical presentation of the structures were performed with the programs RasMol V2.6 (24) and SYBYL 6.5 (Tripos Ass.) (25), respectively. The coordinates were deposited in the Protein Data Bank (entry 1HJI).

RESULTS AND DISCUSSION
One-dimensional NMR Spectroscopy-We have monitored complex formation between the Nun-(10 -44) peptide and the 15 nucleotide boxB RNA by one-dimensional NMR spectroscopy. Free boxB RNA shows three imino proton resonances for G12, G13, and G14. Upon addition of Nun-(10 -44) changes in chemical shifts of these resonances were observed, and two additional resonances for the imino protons of U5 and G6 were detected (Fig. 1). The pattern of resonances is identical to that observed for boxB RNA in complex with N peptides (11,12), suggesting that boxB RNA adopts virtually identical conformations in the presence of either peptide. Spectra of P22 boxB RNA in complex with a P22 N peptide also exhibit an identical pattern of imino proton resonances (10). Structure comparison shows that and P22 boxB RNA are very similar in complex with their respective N peptides, with one base looped out of the apical pentaloop, allowing formation of a typical GNRA tetraloop structure (10 -12). The indole NH resonance of the only Trp residue of N-(1-36), Trp-18, is shifted upfield by more than 1.0 ppm (12) in presence of boxB RNA due to its stacking with the aromatic ring of A7, whereas the indole NH resonance of the single Trp residue of Nun-(10 -44), Trp-33, is shifted upfield by only 0.2 ppm upon boxB RNA complex formation. The comparatively small change in chemical shift for Trp-33 in Nun-(10 -44) may be attributed to conformational changes within the peptide upon RNA binding and already suggests that Trp-33 does not stack with A7 of boxB RNA. In the spectral range above 9 ppm another new signal is observed, originating from Ser-24 NH. For the N-(1-36)-boxB RNA and the P22 N-boxB RNA complex a corresponding signal is observed originating from Ala-3 (12) and Ala-2 (10), respectively.
Assignment of RNA Protons and Verification of the RNA Fold-In two-dimensional spectra all H6/H8, H1Ј, pyrimidine H5, and cytosine amino protons could be assigned. Pyrimidine H5 and H6 were identified in two-dimensional COSY-spectra, cytidine amino protons could be assigned by intraresidual NOEs to their own H5, and purine H8 and all H1Ј were assigned by comparison with spectra of boxB RNA in complex with N-(1-36). In this way, putative assignments were also possible for some isolated ribose resonances in the loop region, which show characteristic upfield (A7) or downfield (A9) shifts. Adenine H2 could be assigned for A7 and A9, and A11H2 could be identified by the strong NOE to U5H3 (Fig. 3).
NOEs indicating formation of Watson-Crick base pairs were observed for C2:G14, C3:G13, C4:G12, and U5:A11. Comparison of the Nun-(20 -44)-boxB RNA and the N-(1-36)-boxB RNA NOESY spectra in the region of the imino protons show that the imino protons in Nun-(20 -44)-boxB RNA are shifted downfield by about 0.05-0.2 ppm, whereas for all non-exchangeable protons no systematic change of the chemical shifts between the spectra could be observed (Fig. 3). Both spectra show an identical pattern of resonances suggesting virtually identical folds in the stem region.
The connectivities between the ribose and their own as well as their sequentially neighboring base were obtained from H1Ј i -H6/H8 i and H1Ј i -H6/H8 iϩ1 -NOEs (26). As in the spectra of N-(1-36)-boxB RNA, no H1Ј i -H6/H8 iϩ1 -NOE could be observed between A8 and A9. This and unusual downfield shifts of the resonances of A9 in the boxB-RNA complexes of both peptides suggest that A9 is looped out the Nun-(10 -44) complex in a fashion virtually identical to that observed in the N-(1-36)-boxB RNA complex. Spatial vicinity of A8 and A10 as in N-(1-36)-boxB RNA complex is possible, but the corresponding NOEs could not be assigned unambiguously due to frequency degeneration. Finally, extreme upfield shifts of U5H5 and A7H1Ј resonances that were also observed in the N-(1-36)-boxB RNA complex lead to the conclusion that boxB RNA adopts a highly similar structure in complex with either Nun- (20 -44) or N-(1-36).
Assignment of Peptide Protons-For Nun-(10 -44) strong TOCSY and relatively weak NOESY signals were observed for the first 10 residues in the complex, suggesting that this region  is flexible and is not involved in RNA binding. Nun-(10 -44) and Nun-(20 -44) showed identical chemical shifts for the overlapping peptide. Thus, all further experiments were performed using Nun-(20 -44).
For the Nun-(20 -44) peptide-boxB RNA complex the sequence-specific resonance assignments could be performed by standard two-dimensional NMR methods (27). NOESY crosspeaks in the backbone amide-amide region indicated the presence of helical structures for a large part of the RNA bound peptide. Unambiguous assignments of the amino acid side chain protons could be performed for all amino acids except Arg-25, Arg-27, Arg-29, Arg-32, Arg-36, and Lys-35 due to frequency degeneration and missing COSY-cross-peaks in the C ␣ H region. Some arginine C ␤ H could be assigned using the helix-typical d ␣␤ (i, i ϩ 3) cross-peaks.
Nun-(20 -44) Forms a Helix in the Complex-The analysis of the C ␣ H chemical shift index (28) gave first evidence for a helical structure from Ser-24 to Asn-43. For this region, helixtypical (i, i ϩ 3) and (i, i ϩ 4) NOEs could be observed throughout the sequence (Fig. 4), in agreement with the observation of only very weak cross-peaks in the TOCSY-spectra for Leu-22 and subsequent residues, probably due to very small 3 J HN␣ coupling constants.
NOE Assignment and Structure Determination-To resolve ambiguities in NOE assignments, an iterative procedure for structure calculation was performed. Initially, only those NOEs that could be assigned unambiguously were used for molecular dynamics calculations. By verification of the resulting structures additional NOEs could be assigned, for example NOEs from the Trp-33 aromatic side chain (Fig. 5) to either Ile-30 or Ile-37.
Intermolecular NOEs-30 intermolecular NOEs could be identified unambiguously in the NOESY spectra. For the ␤-protons of Ser-24 the same NOEs with C2 and C3 are observed as FIG. 7. a, lowest energy structure of the Nun-(20 -44)-boxB RNA complex. RNA surface, gray; peptide Conolly surface, light blue; RNA, orange; peptide backbone, cyan; peptide residues interacting with the RNA (Ser-24, Arg-28, Tyr-39), magenta; residues forming a hydrophobic surface (Leu-22, Ile-30, Trp-33, Ile-37, Leu-41), yellow. The peptide is bound tightly in the major groove of the RNA forming a bent ␣-helix. Tyr-39 stacks on A7, Ser-24 is bound between C2 and C3, and Arg-28 contacts the bases of C4 and U5. b, as in a: RNA surface, gray; peptide Conolly surface, light blue translucent; peptide backbone, cyan; peptide residues represented as sticks in a are shown in spacefill mode. The complex structure is very compact with no cavities found between molecules. Hydrophobic residues are located at the solvent exposed side opposite to the peptide-RNA interface to form a hydrophobic patch. c, overlay of the 20 structures with the lowest overall energy. RNA, as in a: peptide, cyan; hydrophobic side chains as part of the hydrophobic surface patch, yellow; other hydrophobic side chains, magenta; Asn-43, white. A7, Tyr-39, and Asn-43 exhibit the same structural characteristics as A7, Trp-18, and Asn-22 in the N-(1-36)-boxB RNA complex, with Asn-43 packing upon the Tyr-39 aromatic side chain which, in turn, is stacking upon A7. Residues contacting the RNA and forming the hydrophobic surface, respectively, are very well defined. d, the hydrophobic residues form a structure that resembles a sharp ridge which may serve as a recognition site for additional host factors.
Quality of the Nun-(20 -44) Peptide-boxB RNA Complex Structure-The final structure was calculated based on a total number of 191 intra-and 30 intermolecular distance restraints. A subset of 20 structures with energies lower than 182 kcal⅐mol Ϫ1 out of 120 structures was selected for further analysis. None of these structures showed NOE violations larger than 0.13 Å (Table I.).
The backbone of the peptide bound to boxB RNA is well defined for residues 22-43, with a backbone root mean square deviation (r.m.s.d.) of 0.52 Å. Residues 20, 21, and 44 remain disordered so that inclusion of these residues in the r.m.s.d. calculation increases the backbone r.m.s.d. to 0.94 Å. COOHterminal residues and residues forming the hydrophobic surface show the lowest r.m.s.d./residue (Fig. 6) (Ͻ0.5 Å) indicating that the structure is very well defined in these regions.
Structural Features of the Nun-(20 -44)-boxB RNA complex-PROCHECK analysis of Nun- (20 -44) in complex with boxB RNA shows that in the 20 accepted structures 95.9% of the residues are found in the most favored regions, 3.9% in the allowed regions of the Ramachandran plot. The Nun peptide forms an ␣-helix for residues 24 -43, which is bent at residues Ala-31 and Arg-32. Bending does not require deviations from ideal helix ⌽/⌿ angles. The COOH terminus of the peptide is well defined by contacts between Asn-43 and Tyr-39, reflected by several NOEs between side chain protons of both amino acids (Figs. 5 and 7). In the complex, Tyr-39 stacks tightly on A7 anchoring Nun-(20 -44) to the RNA, and the ␥-amino group of Asn-43 packs on the aromatic ring of Tyr-39 (Fig. 7a). The same structural characteristics are exhibited in the N-(1-36)-boxB RNA complex by A7, Trp-18, and Asn-22 (12).
The NH 2 -terminal residues occupy a well defined position close to C3, reflected by the intermolecular NOEs observed between the bases of both C2 and C3 and Ser-24. Arg28 is in close contact to the bases of C4 and U5, which is shown by several NOEs.
According to the NUCPLOT analysis of the 20 converging structures, additional residues critical for the RNA recognition are Arg-27, Arg-32, Arg-36, and Lys-35, which are involved, as well as Arg-28, in numerous electrostatic and hydrophobic contacts ( Fig. 8) with the phosphate backbone and the sugar moieties. In several structures, hydrophobic contacts between the looped out base of A9 and Arg-36 are observed. Arg-36 occupies the position of Gln-15 in N contacting the bases of A7 and A8 in the N-(1-36)-boxB RNA complex.
The most distinguishing feature of the Nun-(20 -44)-boxB RNA structure is the formation of a hydrophobic, solvent-exposed surface. The wheel representation of the Nun-(20 -44) helix in the boxB RNA complex clearly indicates that this helix is amphipathic, and all hydrophobic residues with large side chains are exposed to one side. Corresponding residues of N-(1-36) are either charged or alanines (Fig. 9).
Various NOEs have been assigned between Trp33 aromatic side chain protons and aliphatic protons of Ile-30 and Ile-37. In the D 2 O-NOESY spectra, an NOE between methyl protons of Leu-41 and Ile-37 could be observed. Additionally, the Leu-22 side chain is in spatial vicinity to Ile-30 and also contributes to the hydrophobic surface (Fig. 7b). All these residues are well defined in the structure calculations as shown in the overlay of 20 structures with the lowest overall energy (Fig. 7c). The network of hydrophobic residues forms a clear cut ridge (Fig.  7d) suggesting that the N-boxB-and the Nun-boxB-RNA complexes are recognized by different mechanisms by the target host cell factors.
It has been shown that NusA binds to the COOH-terminal region of Nun and that NusA alone without the other Nus factors inactivates Nun function as a terminator (7). NusA binding to boxB RNA can be nearly abolished by mutation of looped out nucleotide A9 (29). Together with N protein, NusA strongly enhances antitermination, which is also strongly reduced in vivo by mutations of A9 (30) implying that A9 plays a crucial role in antitermination. The current work thus may serve as a basis for the design of specific peptide and RNA mutants aimed at a better understanding of the crucial termi-  FIG. 9. Helical wheel representation of N-(3-20) (a) and Nun-(24 -41) (b). In b, hydrophobic residues with large side chains are in gray. They are all exposed at one side. Corresponding residues in a are either charged or alanines. nation/antitermination regulation of the viral replication on a structural level.