Gottfried Otting (The Australian National University, Australia)
Abstract: Fluorine is bigger than hydrogen and the C-F bond is longer than a C-H bond, but not by much. CF groups prefer hydrophobic environments (think of Teflon). 19F-spins provide site-specific probes easy to detect by 1D 19F-NMR. Using cell-free protein synthesis, we replaced all valine residues in the protein GB1 by fluorinated analogues with a 19F spin in either the CG1 methyl group, the CG2 methyl group or both. The 19F NMR signals were distributed over a large chemical shift range. The protein structure remains unchanged. While CH3 groups rotate rapidly, the CH2F groups preferentially populate different staggered rotamers. Transient contacts between different fluorinated valine residues are manifested by through-space 19F-19F couplings that are observed more easily than 19F-19F NOEs.
-
Hello! An amazing work! Thanks for sharing.
I have a couple of questions.
1) this might be a bit naif but I was wondering: I can clearly appreciate 3 distinct peaks for the 19F-G1 spectrum as well as 4 peaks in the 19F-G2 spectrum. Thus I was expecting 8 peaks in the G1-G2 spectrum. Could you comment on the minor forms and on the relative intensities of the major form of this latter spectrum?
2) You mentioned you were decoupling 1H during 19F acquisitions thus I am assuming you are using a QCI-19F probe (or something similar). If this is the case, did you try any 19F-1H correlation experiment?
3) In the abstract you mention that the protein structure is unchanged upon incorporation of the 19F moiety. How did you prove it? Do you think this would be true even for a putative CF3 group?Thanks again!
-
good questions!
1) Difluorovaline is not as easily accepted by the E. coli valyl-tRNA synthetase as monofluorovaline. Therefore, the tiny amounts of canonical valine present in the cell-free reaction mixture get used preferentially and some of the protein ends up with 3 difluorovalines and 1 valine. This species produces different 19F chemical shifts. Statistically, ~20% of each site contains valine instead of difluorovaline.
2) We use a 400, where 19F is on the X-channel like all other non-1H nuclei. Indeed, to assign the 19F-NMR spectrum, we used 1H-19F correlation spectra.
3) We assigned the 1H NMR spectra. The 1H chemical shifts and NOEs are conserved. Circular dichroism indicates that the melting temperature dropped by ~10 degrees. A CF3 group would perturb the structure more. More critically, it could be quite a challenge for the valyl-tRNA synthetase. -
Hello,
Very interesting and very nice presentation!
Just a few questions:
1) the g1,g2 1D spectrum looks quite different from a ‘visual sum’ of the g1 and g2 1D spectra. Are the CSPs arising from the presence of more 19F in the g1,g2 sample?
2) Did you measure some proton relaxation rates? Relaxation in CH3 (and even more in CF3) methyl groups is quite tricky to analyze, but maybe just the magnitude of the decay would be quite informative on the increased rigidity of the CF3.Cheers
-
1) Indeed, the 19F chemical shifts depend very much on whether there is another fluorine nearby, either in the same amino acid residue or simply in another residue nearby! Based on 1H-1H NOEs and the appearance of Halpha-Hbeta COSY-cross-peaks (reflecting large or small 3J(Alpha,Hbeta) coupling constants), the fluorovaline side chains feature the same Chi1 angles as the valine residues in the wild-type protein. Using a 1H,19F-HOESY spectrum, we obtained stereospecific resonance assignments of the 19F spins in the difluorovaline residues. In 3 of the 4 difluorovaline residues, the relative 19F-chemical shifts (high-field or low-field) proved to be conserved between the samples made with singly fluorinated valines and the sample made with difluorovaline. (Subscripts in the FF-TOCSY spectrum indicate the stereospecific resonance assignments.)
2) Interesting idea! No, we haven’t measured the 1H relaxation of the CH2F groups. (We worked only with CH2F groups, not with CF3 groups, in order to minimise structural perturbations.) Obviously, the 1H relaxation of CH2 groups is difficult to compare with the 1H relaxation of CH3 groups. In an attempt to find evidence for minor rotamer species of the CH2F groups that may be in slow exchange with other rotamers, we performed 19F-CPMG experiments. In the protein made with difluorovaline, only the gamma2-fluorine of residue 54 showed significantly slower relaxation (36 s-1) when we applied 180 degree pulses rapidly as opposed to applying a single refocusing 180 degree pulse (26 s-1). The CH2F group associated with this fluorine atom is right in the hydrophobic core of the protein and more solidly immobilized than the other CH2F groups, which is also demonstrated by a large 3J(1H,19F) coupling. None of the other 19F spins relaxed as quickly.
-
Oops, correction: 26 s-1 with CPMG, 36 s-1 with a single 180(19F) refocussing pulse.
-
Very nice presentation and really fascinating work. I have a somewhat naive question!
Since the CH₂F groups preferentially populate distinct staggered rotamers and exhibit through-space ¹⁹F–¹⁹F couplings, have you explored whether these interactions might also reflect transient conformational states of the protein backbone, rather than being driven purely by side-chain rotamer preferences? And do you think this strategy could be extended to detect low-population backbone conformers that are often invisible to other NMR probes? -
Good thought!
GB1 is a very stable protein and the backbone atoms would not easily deviate far from their average conformations. Nonetheless, in previous work, we found that a through-space scalar 19F-19F coupling can be detected between the CF3 groups of two residues of N6-trifluoroacetyl-L-lysine (TFAK) installed 33 residues apart (one of the TFAK residues being at the C-terminus of a solvent-exposed, flexible polypeptide segment). This observation is interesting because, in this case, the fluorine-fluorine contacts would certainly be transient and infrequent: https://doi.org/10.1021/jacs.1c10104
The big question is, whether a scheme can be designed that uses this effect to detect non-random conformational changes of backbone conformations? I fear that the chemistry may become prohibitive. For example, the alpha-hydrogen would be difficult to replace by fluorine. Furthermore, there would be no detectable scalar coupling, unless the fluorine atoms definitely (and repeatedly) make a contact with some orbital overlap.
Leave a Reply