Hi,
again see below:
One strategy to overcome this was to
* record a HSQC/TROSY at different temperatures at the different
spectrometers
(conveniently by using Brukers multi_zgvt),
Did you use the trosy experiment for relaxation measurements?
Yes, that's why I'm asking. ;)
I have the impression that the methanol method is not always perfectly
reflecting the situation in an aqueous buffer with all kinds of salts in
it, and that the huge methanol signals and the resulting radiation damping
can make determination of the "true" MeOH peak maxima pretty difficult.
They seem pretty broad to me. The maximum resolution of peak distance I can
get with Topspin is 0.02 ppm, and if I consider that 0.02 ppm already makes
a difference of ~ 1 K I guess there is the possibility of significant
error. We tried using deuterated methanol at atmospheric pressure (much
less intense signal, sharper peaks, no boiling point artifacts due to odd
pressures in a sealed tube) to see if there are any differences to our
standard procedure with protonated MeOH, but I couldn't see any.
For the temperature issues, maybe it is worth talking to the Bruker
people. They would know the best. But from my fading memory, I
believe MeOH is useful for temperatures < 300 K whereas ethylene
glycol is for > 300 K. Your issues might arise from methanol
volatility. The different induction and radiation dampening of the
standard samples and your NMR sample is an issue, but there is a lot
of literature on this that might help you. Again this is ancient
memory for me, so you will have to find the articles yourself. There
are many different ways for calibrating the temperature.
To make matters worse, we have four solution spectrometers with Bruker
cryoprobes and one usually used with a room temperature probe, and of
course each and avery probe has a different design in the temperature unit.
It's not hard to start believing that every machine is behaving completely
differently.
Every spectrometer behaves differently with respect to temperature, so
temperature calibration is essential for any serious dynamics study.
It is incorrect and dangerous to assume that the VT temperature
setting is the same as the sample temperature.
To make my point clear I made two figures.
The following plot shows a series of spectra, which all have the same
reference frequency. You can see that most signals are shifting, into
different "directions". One exception is the (alanine) signal at 134 / 9.2
ppm, which is pretty stable over a range of 5 K. The series is color coded
red = 305 K to purple = 310 K according to the usual methanol calibration.
https://dl.dropbox.com/u/4019316/temp-750-zoom.pdf
Now after adding another spectrum from a different spectrometer (in
magenta, it's also a different sample in this case) I first have the
problem of proper referencing, as I don't know if the diverging signal
positions are a result of slightly inaccurate field calibrations or due to
different temperatures. I referenced it to the said Ala signal which seems
to be unimpressed by temperature changes.
The "magenta spectrum" should correspond to one of the spectra "in the
middle", namely 307 K, but in reality it fits the 310 K spectrum much, much
better.
https://dl.dropbox.com/u/4019316/temp-750-reference.pdf
What do you think? Isn't it odd that the temperature-calibrated spectra
don't fit 100%? They should, and the actual sample seems like a better
temperature indicator to me than a somewhat artificial MeOH sample.
Currently I'm measuring a whole set of "TROSY-calibrated" spectra, I'll see
if they give me different results in the consistency tests.
For permanent reference I have attached the two PDF files you have
uploaded to dropbox into a special 'patch' tracker entry at
https://gna.org/patch/?3587. The new file links are:
https://gna.org/patch/download.php?file_id=16912
https://gna.org/patch/download.php?file_id=16913
From your spectra (the above file links), it looks suspiciously as
though you have an exchange process occurring (though it may not be
the case). Have you measured relaxation dispersion data for the
system?
* select the temperature where the NH spectra are nearly 100% identical
If temperature calibration is important, then you should run the
experiment on methanol or ethylene glycol to calibrate. Otherwise if
the spectra are the same but just shifted because of temperature, then
it's not so important.
I never saw any differences in MeOH proton peak distance when running the
different T1/T2/HetNOE experiments against a standard methanol sample,
regardless if it was HSQC-based or TROSY-based.
Maybe MeOH was not suitable for the temperature used?
If you take your NH spectra from different, MeOH-calibrated
magnets/temperature units and superimpose them – are they completely
identical? Don't you see any differences? I always have to re-adjust my
reference peak lists to find the peaks in the spectra of the different
spectrometers.
For me, the spectra are perfectly identical. The only differences are
due to spectral noise shifting the peak randomly in height and
chemical shift in all directions equally.
run a quick 1D after running a short version of the experiment.
That's what I did. I pulsed for 15-20 minutes for the system to equilibrate
and immediatley took a proton 1D to determine the peak distance.
Maybe I should consider using ethylen glycol since the methanol calibration
method seems to be less suited for temperatures around 300-310 K where
we're working with? Are you determining the distance "by hand" or via a
peak-picking mechanism inside topspin? Do you use deuterated methanol or
standard protonated one?
I would recommend ethylene glycol. Have a look at the published
literature on NMR temperature calibration and see what is best. Maybe
you need to modify the phases on your pulses to preserve the ethylene
glycol magnetisation but destroying the water and protein signals, and
then use the data from the recorded 2D for your calibrations. This
would avoid quick cooling between the experiments.
Regards,
Edward
Re: Temperature calibration: methanol- vs protein-based,