Domain-domain motions in proteins from time-modulated pseudocontact shifts.

In recent years paramagnetic NMR derived structural constraints have become increasingly popular for the study of biomolecules. Some of these are based on the distance and angular dependences of pseudo contact shifts (PCSs). When modulated by internal motions PCSs also become sensitive reporters on molecular dynamics. We present here an investigation of the domain-domain motion in a two domain protein (PA0128) through time-modulation of PCSs. PA0128 is a protein of unknown function from Pseudomonas aeruginosa (PA) and contains a Zn(2+) binding site in the N-terminal domain. When substituted with Co(2+) in the binding site, several resonances from the C-terminal domain showed severe line broadening along the (15)N dimension. Relaxation compensated CPMG experiments revealed that the dramatic increase in the (15)N linewidth came from contributions of chemical exchange. Since several sites with perturbed relaxation are localized to a single beta-strand region, and since extracted timescales of motion for the perturbed sites are identical, and since the magnitude of the chemical exchange contributions is consistent with PCSs, the observed rate enhancements are interpreted as the result of concerted domain motion on the timescale of a few milliseconds. Given the predictability of PCS differences and the easy interpretation of the experimental results, we suggest that these effects might be useful in the study of molecular processes occurring on the millisecond to microsecond timescale.

The solution structure of bacteriophage lambda protein W, a small morphogenetic protein possessing a novel fold.

Protein W (gpW) from bacteriophage lambda is required for the stabilization of DNA within the phage head and for attachment of tails onto the head during morphogenesis. Although comprised of only 68 residues, it likely interacts with at least two other proteins in the mature phage and with DNA. Thus, gpW is an intriguing subject for detailed structural studies. We have determined its solution structure using NMR spectroscopy and have found it to possesses a novel fold consisting of two alpha-helices and a single two-stranded beta-sheet arranged around a well-packed hydrophobic core. The 14 C-terminal residues of gpW, which are essential for function, are unstructured in solution.

Quantitative hydroxyl radical footprinting reveals cooperative interactions between DNA-binding subdomains of PU.1 and IRF4.

Quantitative hydroxyl radical footprinting and fluorescence polarization measurements have been used to determine the dissociation constants (Kd) of complexes between the ets domain of the murine transcription factor PU.1 and three different DNA fragments. Two natural PU.1 binding sites, the SV40 enhancer site and the lambdaB motif of Iglambda2-4 enhancer, were used as well as the PU.1 binding site present in the crystallized PU.1-DNA complex. With the use of quantitative hydroxyl radical footprinting we obtained binding isotherms for individual protected nucleotides and contact sites on both strands of the DNA. Kd values of (1.53 +/- 0. 12) x 10(-)8 M were found for the lambdaB element, (3.60 +/- 0.65) x 10(-)8 M for the SV40 enhancer site, and (2.28 +/- 0.27) x 10(-)8 M for the sequence used in the crystal structure. In addition, the binding of a second protein, the DNA binding domain of IRF4, to the lambdaB site by itself and in the presence of PU.1 was analyzed. The IRF4 DBD shows three footprints on the TTCC strand and one footprint on the GGAA strand of the lambdaB element. The dissociation constant for the binary IRF4 DBD-lambdaB complex equals (5.59 +/- 0.60) x 10(-)7 M. The Kd value of the IRF4-lambdaB interaction is reduced by a factor of 5 in the presence of two different DNA-bound PU.1 protein constructs, PU.1 DBD and a PU.1 construct containing the PEST domain (PU.1-PEST). A similar decrease of the Kd value was observed for the binding of PU.1-PEST in the presence of DNA-bound IRF4 DBD demonstrating a cooperative interaction between the PU. 1-PEST and IRF4 DBD. On the basis of the hydroxyl radical footprints in the ternary PU.1/IRF4/lambdaB complex, a model for the interactions between the two proteins and the lambdaB site was developed. The DNA binding domains of both proteins bind the DNA in the major groove with potential protein-protein interactions near the intervening minor groove.

Cooperative interaction between the DNA-binding domains of PU.1 and IRF4.

The two lymphoid-specific transcription factors PU.1 and IRF4 form a cooperative ternary complex at immunoglobulin enhancer elements such as the lambdaB and kappaE3' sites. We report here that the synergy of this interaction can be reconstituted in part with the DNA-binding domains of the two proteins. The minimal DNA binding-domain of IRF4 was mapped to residues 20 to 137, corresponding to the conserved DNA-binding region of other interferon regulatory factors (IRFs). This domain can bind weakly to a synthetic murine lambdaB element, while IRF4 constructs that contain residues 1 to 19 require the presence of PU.1 for DNA-binding at similar concentrations. Fluorescence polarization of fluorescein-labelled DNA was used to show that the presence of residues 1 to 19 decreases the binding affinity of IRF4 N-terminal constructs from two- to fivefold. However, all constructs bound better to the lambdaB element in the presence of the DNA-binding domain of PU.1. This cooperative interaction was not dependent on phosphorylation of the PEST domain of PU.1, but was dependent on the proper spacing of the binding sites for PU.1 and IRF4. These data suggest that at least part of the cooperative interaction between full-length PU.1 and IRF4 involves the DNA-binding domains of the two proteins. NMR spectroscopy of 15N-labelled PU.1 and IRF4 constructs indicates that the PEST domain of PU.1 and residues 1 to 19 of IRF4 may be unstructured in the isolated proteins.

The interactions with solvent, heat stability, and 13C-labelling of alamethicin, an ion-channel-forming peptide.

The peptide alamethicin was labelled with 13C and 15N by growing the fungus Trichoderma viride in a medium containing [U-13C] glucose and K15NO3. Spin-echo difference spectroscopy showed that 13C was incorporated to a level of about 50% and 15N to about 98%. Incorporation of 13C into the peptide provided residue-specific probes of the interactions with solvent and heat stability of this ion-channel-forming peptide. All of the carbonyl carbons and the alpha-carbons of the alpha-aminoisobutyric acid [Ala(Me)] residues of alamethicin in methanol were assigned using two-dimensional and three-dimensional heteronuclear correlation experiments. Measurements of 1JC'N revealed hydrogen bonding with solvent at residues 1 and 19 at the ends of the peptide and at Gly11 in the middle. The data also support the thesis [see Juranic, N., Ilich, P. K. & Macara, S. (1995) J. Am. Chem. Soc. 117, 405-410 that intramolecular hydrogen bonds in proteins and peptides are weaker than hydrogen bonds to solvent. The sensitivity of alamethicin carbonyl and proton chemical shifts to perturbation by dimethyl sulfoxide correlates well with the calculated solvent accessibilities of the carbonyls in the crystal structures and reveals residues in the middle of the peptide and at the C-terminus which interact with solvent. Taken together with the 1JC'N measurements, the data support a model in which hydrogen bonding to solvent at the Gly11/Leu12 amide could provide a site of hydration in the interior of the alamethicin channel structure. The temperature dependencies of the carbonyl chemical shifts support the suggestion that the peptide is flexible in the regions where solvent interacts with the backbone of the peptide. The linear temperature dependence of the carbonyl chemical shifts and molar ellipticity indicate that, due to steric constraints at the Ala(Me) residues, the peptide folding/unfolding transition is non-cooperative and that the peptide is remarkably heat stable.

Backbone dynamics of an alamethicin in methanol and aqueous detergent solution determined by heteronuclear (1)H- (15)N NMR spectroscopy.

The (15)N relaxation rates of the α-aminoisobutyric acid (Aib)-rich peptide alamethicin dissolved in methanol at 27°C and 5°C, and dissolved in aqueous sodium dodecylsulfate (SDS) at 27°C, were measured using inverse-detected one-and two-dimensional (1)H-(15)N NMR spectroscopy. Measurements of (15)N longitudinal (R(N)(N(z))) and transverse (R(N)(N(x,y))) relaxation rates and the {(1)H} (15)N nuclear Overhauser enhancement (NOE) at 11.7 Tesla were used to calculate (quasi-) spectral density values at 0, 50, and 450 MHz for the peptide in methanol and in SDS. Spectral density mapping at 0, 50, 450, 500, and 550 MHz was done using additional measurements of the (1)H-(15)N lingitudinal two-spin order, R(NH)(2H (infZ) (supN) N(Z)), two-spin antiphase coherence, R(NH)(2H (infN) (supZ) N(x,y)), and the proton longitudinal relaxation rate, R(H)(H (infN) (supZ) ), for the peptide dissolved in methanol only. The spectral density of motions was also modeled using the three-parameter Lipari-Szabo function. The overall rotational correlation times were determined to be 1.1, 2.5, and 5.7 ns for alamethicin in methanol at 27°C and 5°C, and in SDS at 27°C, respectively. From the rotational correlation time determined in SDS the number of detergent molecules associated with the peptide was estimated to be about 40. The average order parameter was about 0.7 and the internal correlation times were about 70 ps for the majority of backbone amide (15)N sites of alamethicin in methanol and in SDS. The relaxation data, spectral densities, and order parameters suggest that the peptide N-H vectors of alamethicin are not as highly constrained as the 'core' regions of folded globular proteins. However, the peptide backbone is clearly not as mobile as the most unconstrained regions of folded proteins, such as those found in the 'frayed' C-and N-termini of some proteins, or in randomcoil peptides. The data also suggest significant mobility at both ends of the peptide dissolved in methanol. In SDS the mobility in the middle and at the ends of the peptide is reduced. The implications of the results with respect to the sterically hindered Aib residues and the biological activities of the peptide are discussed.

Uniform 15N labeling of a fungal peptide: the structure and dynamics of an alamethicin by 15N and 1H NMR spectroscopy.

An alamethicin, secreted by the fungus Trichoderma viride and containing a glutamine at position 18 instead of the usual glutamic acid, has been uniformly labeled with 15N and purified by HPLC. The extent of 15N incorporation at individual backbone and side-chain sites was found to vary from 85% to 92%, as measured by spin-echo difference spectroscopy. The proton NMR spectrum of the peptide dissolved in methanol was assigned using correlation spectroscopies and nuclear Overhauser enhancements (NOE) measured in the rotating frame. The 15N resonances were assigned by the 2D 1H-15N correlation via heteronuclear multiple-quantum coherence experiment. NOEs and 3JNHC alpha H coupling constants strongly suggest that, in methanol, from Aib-3 to Gly-11, the peptide adopts a predominantly helical conformation, in agreement with previous 1H NMR studies [Esposito, G., Carver, J.A, Boyd, J., & Campbell, I.D. (1987) Biochemistry 26, 1043-1050; Banerjee, U., Tsui, F.-P., Balasubramanian, T.N., Marshall, G.R., & Chan, S I. (1983) J. Mol. Biol. 165, 757-775]. The conformation of the carboxyl terminus (12-20) is less well determined, partly because the amino acid composition reduces the number of NOEs and coupling constants which can be determined by 1H NMR spectroscopy. The 3JNHC alpha H in the C-terminus suggest the possibility of conformational averaging at Leu-12, Val-15, and Gln-19, an interpretation which is supported by a recent molecular dynamics simulation of the peptide [Fraternalli, F. (1990) Biopolymers 30, 1083-1099].(ABSTRACT TRUNCATED AT 250 WORDS)