Structure‐function relationship in Escherichia coli translational initiation factors Characterization of IF1 by high‐resolution 1H‐NMR spectroscopy

Escherichia coli translational initiation factor IF1 was studied by 1H‐NMR spectroscopy at 400 MHz. IF1 displays a very well resolved spectrum in both aromatic and aliphatic regions. Other spectral characteristics include relatively narrow resonance lines and lack of relevant cross‐relaxation phenomena. The resonances of the aromatic residues, in particular of the two His and two Tyr, were assigned by selective chemical modifications and spectroscopic techniques to individual residues in the protein sequence. The relative mobility of various residues of IF1 has been evaluated on the basis of the spin‐lattice relaxation times which are rather short and homogeneous. Overall the factor appears to have a complex secondary and tertiary structure and to be a flexible protein whose residues have a high degree of internal mobility.


INTRODUCTION
Initiation factor IF1 (Mr 8119), the smallest of the three protein factors required for translational initiation in Escherichia coli (reviews [1,2]), consists of 71 amino acid redidues of known sequence [3]. A gene coding for a protein with similar size and homologous primary structure has recently been found in the chloroplast DNA of two distant plants [4,5]. During protein synthesis, IF1 binds, presumably in combination with the other two initiation factors, to the free 30 S ribosomal subunits [6]. Omission of IF1 from reaction mixtures containing all purified components required for protein synthesis results in a much reduced translational rate [7]. This effect can be explained by the kinetic role of IF1 in accelerating the rate of formation of the 30 S initiation complex [8,9].
Correspondence address: C.O. Gualerzi Upon the joining of a 50 S ribosomal subunit to the 30 S initiation complex, which generates the 70 S initiation complex, IF1 is ejected from the 30 S particle [6].
A better understanding of the mode of action of IF1 awaits knowledge of its structure and active sites. A step towards this end is the characterization of IF1 by high-resolution NMR spectroscopy and the assignment of its aromatic spectrum which are presented in this paper.

MATERIALS AND METHODS
Purified, electrophoretically homogeneous IF1 was prepared from E. coli MRE600 cells according to [lo].
NMR spectra were run with a Bruker NMR WH-400 spectrometer using a 30" pulsewidth.
Free induction decays (FIDs) were collected on 16 K memory for lOOl-2000 scans and then Fourier transformed.
Unless otherwise specified, no filtering was used. The dependence of chemical shift on pH was measured following addition of minute amounts of DCI or NaOD to samples. The pH was measured in the NMR tubes with a microelectrode and meter readings were used without further correction. Double-resonance  experiments  were performed  using standard homo-gated  decoupling techniques.  Spin-lattice  relaxation  times were measured  by the inversion recovery T-180-7-90-FID pulse sequence; T (waiting time between sequences) was chosen to be sufficiently long to allow the spin system to relax completely after the acquisition.
To observe the relaxation behavior and to measure the related time constants of the individual resonances (rt), the 7 value ranged from about 1 s to a few tenths of a millisecond.

The 400 MHz 'H-NMR spectrum of IFI
The spectrum of IF1 is very well resolved in both aromatic and aliphatic regions ( fig.1). As shown below, most of the aromatic resonances could be assigned by spectroscopic and protein chemical methods to specific amino acid residues. In the aliphatic region, however, due to the complexity of the spectrum which contains several bands, single resonances and a discrete number of ring-current shifted methyl (RCSM) resonances, only a few tentative assignments [l 1,121 could be made.

pH dependence of the 'H-NMR spectrum
The pH range amenable for investigation of IF1 is pH 5.5-8.5; above or below these values and at the protein concentrations necessary for NMR spectroscopy, precipitation occurs. Within this pH range, the C-2 and C-4 protons of the two histidines of the molecule undergo substantial changes in chemical shift, unaccompanied by changes in other regions of the spectrum.
The titration of the two C-2 protons is presented in fig.2A,B.
The log(6,-bo~s)/(&bs -&) vs pH plots are linear for both His residues, indicating that, at least within the pH range investigated, there are no co-titrating groups. The measured pK values were 6.68 + 0.05 for His A ( fig.2A) and 6.35 f 0.05 for His B ( fig.2B). Unlike the case with the C-2 protons, the chemical shift changes of the C-4 proton resonances cannot be followed throughout the pH range due to partial or complete overlapping with other resonances.

Assignment of the His resonances
IF1 contains two His residues at positions 29 and 34 of its primary structure [3]. Owing to the difference in their pK values (cf. fig.2A,B), assignment of their resonances in the spectrum was obtained following dye-sensitized photooxidation at different pH values. Since photooxidation takes place on the deprotonated imidazole ring and since  we found that His 34 becomes photooxidized at lower pH than His 29 [ 131, we conclude that His 34 has a lower pK than His 29 and therefore its C-2 resonance corresponds to that denoted His B in the spectrum. Consequently, the C-2 resonance of His 29 must correspond to that denoted His A. The assignment of the C-4 protons of these residues was made possible by the inversion recovery experiment presented below.
The values of the spin-lattice relaxation times obtained by the inversion recovery pulse sequence (see below) indicate that the two His residues or at least their aromatic rings are engaged to different extents in the tertiary structure of the molecule. In fact, the resonances due to the protons of the imidazole ring of His 34 show relaxation times shorter than those measured for the ring protons of His 29, indicating that the latter residue has a higher degree of mobility than His 34. The lower pK value found for the imidazole protons of His 34 also suggests that this residue is most likely involved in hydrogen bonding within the tertiary structure of the protein.

Assignment of the Tyr resonances
The spin systems of the 2,6 and 3,5 protons of the two Tyr residues have been identified by homo- nuclear decoupling (fig.3). The assignment of these residues within the primary structure of the protein was obtained by selective chemical modification experiments.
The method used for identification of the Tyr resonances (i.e. comparison of the spectra recorded before and after nitration of tyrosines with tetranitromethane) has been successfully applied in other cases [14].
The spectrum of the modified protein ( fig.4b) shows 'new' resonances (stippled areas) which partly overlap those of the native protein ( fig.4a). The new resonances are due to the protons of the nitro-substituted aromatic rings. In fact, nitration causes the disruption of the symmetry of the AA'BB' spin system of the Tyr ring with the establishment of two ABC spin systems depending on which carbon is nitrated.
The precise assignment of the resonances arising from the nitration of the protein is beyond the scope of this work.
However, since the two Tyr residues of IF1 (position 43,59) react at different rates [15], we were able to assign the individual resonances to specific Tyr residues by comparing the spectra recorded after different times of reaction. For this purpose, we performed the deconvolution with a Lorentzian An example of a computer-generated spectrum obtained as a sum of the deconvoluted resonances is presented in fig.4c. The decrease in integrated intensities of the 3,5 protons of the two Tyr residues as a function of reaction time was followed upon normalization of the resonances (fig.5). As seen from the figure, the resonance intensities of the two Tyr residues decrease at different rates, the one labelled 'A' decreasing faster than 'B'. Since it has been found that Tyr 59 is more accessible to chemical modification than Tyr 43 [15], we conclude that Tyr A and Tyr B correspond to Tyr 59 and Tyr 43, respectively.

Spin-lattice relaxation times
The spin-lattice relaxation times (Tt) for individual proton resonances and bands were determined by inversion recovery. The stacked plot of the aromatic region of this experiment from which we were also able to assign the C-4 proton resonances of the two His residues is presented in fig.6 for all the amino acid redisues, suggesting that the protein is endowed with a rather high degree of internal mobility. The resonance of His B (His 34) however, displays a much shorter relaxation time than His A (His 29), as is also evident from the spectra shown in fig.6. This behavior indicates that His 34 is characterized by a much lower degree of internal mobility than His 29.

CONCLUSIONS
A 'H-NMR spectroscopic study of translational initiation factor IF1 has been presented. Several spectral features such as the abundant ring current shifted methyl resonances and the evidence of clearly distinct chemical and magnetic environments for the two His and Tyr residues of the protein indicate that, in spite of its size, IF1 is endowed with a rather complex secondary and tertiary structure.
The size of this protein, the good resolution achieved and the considerable spreading of its resonances in both aromatic and aliphatic regions of the spectrum as well as the rather narrow lineshapes suggest that the application of 2D NMR spectroscopic techniques to this protein may eventually lead to the elucidation of its 3D structure in solution.
The recent assembly of a synthetic gene encoding IF1 designed to allow facile cassette mutagenesis and the very large amount of IF1 obtainable from an overproducing strain carrying this gene [17] should facilitate this task. Furthermore, site-directed mutagenesis, combined with the 2D NMR technique, may not only represent a method for checking the spectral assignments, but may also offer the opportunity to investigate the Table I TI effect of single amino acid replacements on the secondary and tertiary structure. These correlations, in combination with data on the biological effects of these manipulations, can ultimately yield important information concerning the relationships between structure and function of this factor.