HNCACBSE planes.

6/30/02 - Steve Hardies

Background and purpose.

The HNCACB experiment is a 3D determination that takes several days of machine time. In order to see in advance that the pulse times, etc., are OK, and that the sample intensity is OK, one should run two preliminary experiments called the Carbon and Nitrogen planes. These are run by commenting out one of the following lines in the pulse program:

#define N15_EVOL ; comment out for 2D w/o N15 evolution
;#define C13_EVOL ; comment out for 2D w/o C13 evolution

As shown above, 13C evolution is commented out causing a Nitrogen plane to be acquired. For samples of good intensity (ie. comparable to the ubiquitin control) these determinations can be made in about 30 min for the N15 plane and about an hour for the C13 plane (on the Avance600). The result is a 2D ¹H x ¹⁵N or ¹H x ¹³C spectrum, which will correspond to the projection onto that respective plane of the full 3D data set when it is collected. Generally one puts these in two different experiment numbers in the same database.
[I suspect that such a projection can be computed from the 3D data set for comparison, but I don't know if it customarily done, and these instructions do not address how to do it.]

NOTE: All of the pulse times should be set for each of the planes determinations.

Correspondence to first serial file check.

As with the HSQC, it is possible to check the first serial file as a 1D proton spectrum to detect impending catistrophic failure of the acquisition. However, unlike the HSQC, the first serial file produces a rather indistinct spectrum. Typically, the entire envelope of the amide region signal is only about 3x the baseline noise and without distinct peaks. For the carbon plane, the phase of the signals in the first serial file are all positive, so the signal can be phased to bulge upwards as an indistinct envelope. For the Nitrogen plane, the first serial file has both positive and negative signals. So the amide region appears as about a 3x thickening of the baseline noise. For Nitrogen planes, the signal is stronger in later serial files. For example, try 5 rser, or 6 rser. Failure to see the signal described above at the serial file level would indicate that the planes acquisition should be stopped and the setup proofed for typographical or other errors. However the indistinctness of the spectrum at the serial file level recommends for the collection and processing of the planes data to detect and correct more subtle problems before committing the machine for several days.

[Having said that, I lack any quantitative criteria with which to judge the planes data.]

Expected appearance of the planes data:

While learning, the Nitrogen plane should be processed first. It looks a lot like the ¹H - ¹⁵N HSQC. Some signals will be missing. Others may be folded in as negative peaks due to the generally tighter ¹⁵N spectral width. The Carbon planes data has both positive and negative signals in the final processed spectrum. They are distributed at the same ¹H positions. However, each 1H resonance generates more than one peak in the ¹³C dimension. Typically there is a strong positive and a strong negative peak vertically aligned for each amide proton detected. Looking more closely (which may require lowering the first contour level and admitting some noise) there may be additional weak positive and weak negative signals also vertically aligned with the stronger peaks. It helps to note the chemical shifts of the more isolated signals in the Nitrogen plane, and then look at those positions in the Carbon plane. The positive signals in the Carbon plane tend to fall between 45-65 ppm. Negative signals tend to fall between 30-45 ppm. There may also be negave signals in the 65-75ppm range. Some of the negative signals probably originate from folded peaks.

Sample data:

I've set aside some processed data, and the corresponding datasets and processing scripts for the ubiquitin standard.
In the NIS system, go to /instinct2/hardies/standards
The hncacbse.ref dataset, which I believe was acquired by Andy, is processed there. Experno 1 is the Carbon plane, and Experno2 is the Nitrogen plane. In experno1, comp.ft2 is the finished 2D spectrum properly processed (or at least to the best of my current ability) and it can be loaded into nmrDraw and viewed. For experno 1, test.ft2 shows what happens if you leave the MODE for the y axis as Echo-antiEcho, rather than changing it to Complex. The other dataset, hsqc_fb.ubi, was copied from one of Andy's directories. It shows the ubquitin ¹H - ¹⁵N HSQC in test.dat.

Processing.

The processing described here is by nmrPipe in the NIS system.
[I don't know how to do it in XwinNMR. That's still on the list of things to figure out. I suspect that's how most people do it.]
Processing generally follows along the lines of an HSQC processing with the following major differences:

If you are confused about which experno was the N plane and which was the C plane, use vi to view the file pulseprogram in the copied dataset/experno. The define statement that was commented out will still appear in this preprocessed version of the pulse program as a comment. The define statement that was actually executed will have disappeared.
If the bruker conversion program puts Echo-antiEcho in for any of the MODEs, then that must be changed to Complex. Note in the sample data above the disaster that ensues otherwise.

Otherwise the bruker script appears to work correctly.
You should check the carrier positions, but they are presumably correctly read off of sfo1,2,3.
2nd dimension SW for both planes is 1/2*inr0, as correctly reported by the Bruker script.
[Although the pulse program comments refer to zMODE, when collecting the 2D planes data, the X atom is always y. yMODE should always be Complex]

The appearance after the first Fourier transform is much like the first serial file (a lot of baseline noise with signal only about 3x noise and no distinct peaks). Manually phasing at this point doesn't get you very close. Readjust the phase after getting the full 2D transformed data. There's also probably not much point in messing around with baseline corrections at this step, since the baseline is so noisey.
After the first dimension processing steps (including TP) setup just like for HSQC (including adding SOL, and EXT), add the second dimension processing steps recommeded in the pulse program. Then fine tune to taste. Note that there are different 2nd dimension processing steps for Carbon planes and Nitrogen planes. There are also different steps required for Carbon plane processing depending on whether cnst0 was set as 0 or 1.
[Until I find criteria for using this data, I won't know how much fine tuning is justified.]

Comments from HNCACBSE.ref pulse program (version 6/3/02) about processing:

Note: Always check the extant ref program for updated comments.

;15N Dimension Processing
;Conversion: Dim = y, zMODE = Complex, aq2D = States
;Processing:
;|nmrPipe -fn SP -off 0.45 -end 0.98 -pow 1 -c 0.5    \
;|nmrPipe -fn ZF -size 128                             \
;|nmrPipe -fn FT -neg                                  \
;|nmrPipe -fn PS -p0 0.0 -p1 0.0 -di -verb             \
;|nmrPipe -fn POLY -auto -ord 1                        \

;13C Dimension Processing
;Conversion: Dim = z, yMODE = Complex, aq2D = States
;Processing (depends on the value of cnst0; shown below is example for cnst0 = 1)
;|nmrPipe -fn ZF -pad 1                                \
;|nmrPipe -fn RS -rs 1 -sw                             \
;|nmrPipe -fn LP -before -pred 1                       \
;|nmrPipe -fn SP -off 0.5 -end 0.98 -pow 2 -c 1.0     \
;|nmrPipe -fn ZF -size 512                             \
;|nmrPipe -fn FT                                       \
;|nmrPipe -fn PS -p0 -90.0 -p1 180.0 -di -verb         \

;Processing (depends on the value of cnst0; shown below is example for cnst0 = 0)
;|nmrPipe -fn SP -off 0.5 -end 0.98 -pow 2 -c 1.0     \
;|nmrPipe -fn ZF -size 512                             \
;|nmrPipe -fn FT                                       \
;|nmrPipe -fn PS -p0 -90.0 -p1 180.0 -di -verb         \

;Testing
;C13 plane gives the strongest signal since N15 undergoes 0/0 phase shift
;for the first N15 point

;N15 plane gives a weak signal since Ca/Cb signals are of opposite sign and
;tend to cancel. Additionally, C13 undergoes either -90/180 or -270/540
;phase shift for cnst0=0 and cnst0=1, respectively, which causes significant
;dephasing for the first fid

Comments:

Because of the high noise levels, SP -off = 0.45 or 0.4 or even lower may be better for all cases.
a POLY -auto -ord 1 after the last phasing may also be useful in the other cases, or at least I did it and it didn't hurt.
[I assume Dim = y and Dim = z are not bruker switches, but just telling you which atom will be on which axis when you collect 3D data. I don't understand why zMODE is mentioned after Dim=y and vice versa.]
The comments about testing refer to the appearance of the first serial file.

Interpreting HNCACB planes data:

The Nitrogen plane should look like the HSQC, except some peaks will be missing, and the folding pattern will be different because the spectral width is different. The peak intensity is much more sensitive to T2 relaxation than an HSQC, because the pulse sequence cycle takes longer to complete. Peaks from disordered regions will be greatly exaggerated in intensity. At NS=32 for the plane, if you can see even a faint spot for most of the other peaks, then the 3D acquisition at NS=16 will be OK. If you can't see a lot of the other peaks, then the 3D acquisition at NS=16 may still be OK. (These values for the Avance600). NS=16 for the 3D determination is 3 days, so we don't really ever try to extend NS beyond 24. When the 700 comes on line, it may be used to ramp up the sensitivity.
Note: disordered regions also make the rser 1 spectrum look more impressive, but this is not particularly a good thing.
Low signal intensity is because the protein is too big and rotating too slowly. It seems to me we ought to be able to directly calculate from the T2N15 measurements which peaks will be unacceptably faint and forego the HNCACB if there are many.
If HNCACB is too faint, then there are two other experiments that can be done that are more tolerant of a fast T2 and that give the same data.
The Carbon plane should ideally have 4 signals lined up on the 1H coordinate for each signal in the Nitrogen plane (with a few exceptions). Two should be positive and two should be negative. Generally one of the positive and one of the negative signals should be stronger than the other. The same intensity issues apply as for the N plane. So, basically, I know of no information that the C plane would give that would override the decision made to go on from the N plane.
The CBCACONH experiment that follows and reinforces the HNCACB gives about 3 x greater intensity than the HNCACB according to a figure in the Bruker 3D acquisition manual. Actually, this should depend on the T2 relaxation time. If the CBCACONH is faster to cycle through its pulse program, once again we should be able to calculate in advance the range of T2 times that will be tolerated.

Issues:

It seems to me that the available criteria for using the planes data could be improved.

It seems that we ought to know from the T2N15 some peaks that should be marginally visible, and if they are not we ought to have criteria to say if there is a problem with signal intensity, or if the noise is too high. Then we ought to have a list of things to recheck in either of those cases to fix it.

Is it true that you can compute the planes from 3D data, and is there any reason to do this? Would you exepct such data to be of the same quality as the 2D-acquired planes, or would it be superior?
Should the Nitrogen plane look exactly like the 1H - 15N HSQC except for some missing peaks and some folded peaks due to ¹⁵N spectral width? Should the ¹⁵N spectral width be adjusted to avoid folded peaks for the full HNCACB acquisition? Or is the spectral width tighter for economy of time, and is it true that one can still interpret the folded peaks? Are the negative signals in the 65-75ppm range from folded peaks?
Look up and incorporate which atoms are giving the negative and which the positive signals in the Carbon plane.
What was Dim = x (or y) about in the pulse program. Why is it mentioning zMODE? , particulary why is zMODE mentioned in the context of Dim=y and vice versa?
I had some problem I don't remember having before where the nmrproc.com script would just hang prior to the first phasing step. There were no spaces after any \ marks. Finally by trial and error I got it to work by removing a -ov -out tag series after the first Fourier transform. But I remember being able to write a file there like that before. What does the / really mean? Is it a unix thing or a syntax specific to these programs?