15N and 13C 90 Degree Pulse Time Calibration.

Version 7/2/02   SCH

Purpose:

15N 90 degree pulses are delivered in most of the pulse sequences involved in protein characterization (i.e.. HSQC, HNCACB).  13C 90 degree pulses are delivered in most pulse sequences involving 13C-labeled samples.  13C and 15N  pulse times should be relatively constant for a given spectrometer (field strength), temperature, power level, and type of pulse (square or one of the shaped profiles).  When first setting up a type of experiment, one is strongly encouraged to calibrate the pulse times.    They need not be recalibrated prior to each use of the procedure.  However, if there is a problem with a type of experiment, recalibrating the pulse times is a good way to check out the operation of each channel separately and diagnose a problem.  These calibrations are done with special samples.  If done in conjunction with a protein determination (such as HSQC or HNCACB),  the calibrations should be completed before loading, tuning, shimming with the protein sample.

Reference samples:

The 15N determination is done with 15N-labeled urea in 2H-labeled DMSO.  DMSO is used as the solvent instead of water to avoid chemical exchange of urea protons with the solvent.

The 13C calibration is done with a sample of acetate 13C labeled on carbon 2 (methyl group).  The acetate is dissolved in deuterated water.

Both reference samples may be found in Wilmad tubes (round bottom NMR tubes) found in the NMR room.

Temperature:

The temperature should be set to the same temperature as the protein determination being calibrated.

Data sets:

The data sets and pulse programs with suitable parameters set are named calib15n.ref and calib13c.ref.  Remember to make a copy of both the data set and the pulse program under your own ownership and to make the pulprog parameter in the data set point to your copy of the pulse program.

Deuterium Lock:

The deuterium lock is used for both calibration experiments.  It is set by entering the command lock d2o.  The same command is used for the 15N urea sample, even though it is in deuterated DMSO instead of deuterated water.  This command replaces the more elaborate procedure we previously used to set the deuterium lock on the AMX 500.

Shimming:

The calibration experiments do not require precise shimming.  Look for a shim file to load like ref_31c_acetate_wilmad or ref_15n_urea_wilmad..  Use the manual shimming procedure from the HSQC procedure, but it is sufficient to do it expediently.  The usual gradshim program can not be used because it works on the 1H signal from water, which is not present in either of these samples.   There is a 1 dimensional gradshim routine that works on the deuterium signal named 1D2H that you could alternatively try.

Tuning 15N or 13C probe:

The 15N and 13C channels are not sensitive to changes in the conductivity of the sample (i.e.. salt) the way the 1H channel is.  Tuning of 15N and 13C channels is mostly conducted when trying to identify a problem affecting the setup of an experiment.  Tuning of the 13C probe is analogous to tuning the 1H channel, except a tuning data set called tune13c.ref should be used. The pots to be adjusted  for 15N and 13C are each different, and are color coded as marked on the magnet in the probe tuning area.  Tuning of the 15N probe requires an extra step on the Avance 600.  The pre amplifier unit on the floor by the magnet and having the display used for tuning can only handle only one X channel at a time (i.e. 15N or 13C).  The system is usually set up to route the 13C channel through the pre amplifier.  Thus, for 15N tuning, the 15N signal has to be rerouted through one of the pre amplifier channel usually used for 13C.  This is partly accomplished by parameters set in the tune15n data set, which reroute the RF so that 15N RF enters the rear of the preamplifier instead of 13C RF..  To see how signals are routed by data set parameters use the edasp command and compare the tune15n data set to other data sets.

To reroute the 15N RF through the preamplifier and to the probe in the avance 600 also requires a hardware change.  This involves switching the cable (and accompanying filter) which connects the front of the X-nucleus preamp (the second one down in the stack) to the probe. Normally a 13C bandpass filter is attached to the front of the X-nucleus preamp which in turn is connected to the 13C channel on the probe by a cable. This should be changed so that the 15N bandpass filter is attached to the front of the preamp which in turn is connected to the 15N channel of the probe. Normally the 15N channel is set up so that the input from the amplifier bypasses the preamp, is passed through the 15N bandpass filter, and then into the 15N channel on the probe. Thus, for the purposes of the 15N tuning experiment, the connection between the 15N amplifier and 15N bandpass filter can be broken and in turn the 15N bandpass filter attached to the front of the X-nucleus preamp. It's OK if the 13C connection is broken during the 15N tuning operation). Two important things to remember are to a) Make sure you change the cable back after you tune the 15N probe [including before running calib15n], and b) be very gentle with the 15N and 13C bandpass filters; they contain inductors (essentially small coils of copper wire) that are sensitive to mechanical distortion which in turn will alter their filtering properties.

The Avance 500 and 700 spectrometers will have two X-nucleus preamplifier units each to avoid having to change cables.

Pulse sequence and theory of the experiment.

In this experiment, energy is first put into the coupled spin system with a 90 degree proton pulse.  If one were to take a proton FID immediately and Fourier transform it, the signal originating from the test molecule proton would be split into two equal peaks by scalar (through bond) coupling to the 13C or 15N atom.  The two peaks would be separated by a number of Hz characteristic of this molecule.  The separation is known as 1JHN or 1JHC, about 92 and 140 Hz, respectively.   (The 1 indicates coupling through only 1 bond).  The J coupling constants are field independent, so they will be the same on the Avance 500, 600, and 700 MHz NMR spectrometers.   Rather than immediately collecting the FID as described above, the calibration pulse sequences delay for a time after the 90 degree proton pulse called the "delta delay" or the "dephasing time".  During this time the spin systems of the 1H and the bonded 13C or 15N interact via their scalar or J coupling causing the phase of the two 1H signals to become different.  The delay time in the pulse programs is adjusted to 1/(2J), which is the appropriate amount to cause the peaks to become 180 degrees out of phase.  This pattern is referred to as "antiphase".  When one of the peaks is phased as a positive peak, the other will be negative.  The proton carrier frequency in the data set (SFO1) should have been set to the average (or uncoupled) frequency of the proton, so the two peaks should be at +1/2 J Hz and - 1/2 J Hz away from the carrier position.  In the product operator formalism (see Edison et al., Meth. Enzymol. 239: 3-79 (1994)), the 180 degree dephasing time results in the transition from an Ix state to a 2IySz single quantum state. (I = 1H and S = 13C or 15N).

As soon as the antiphase state is reached, the 90 degree 15N or 13C pulse to be calibrated is delivered.  If the pulse is correct, it will convert the 2IySz state to the multiple quantum, 2IySy state.  This is called a multiple quantum state because there is transverse spin (x or y) on both of the coupled atoms.  In accordance with quantum theory, multiple quantum states do not produce a signal.  The calibration experiment consists of varying the 15N or 13C pulse time to make the antiphase signal disappear.  Pulse times past 90 degrees cause the antiphase signal to reappear with both peaks inverted relative to the 0 pulse time antiphase pattern.  A 180 degree pulse causes the inverted antiphase pattern to reach a maximum intensity.

Carrying out the calibration.

For carbon calibration, make sure the carrier is set to the resonance of the reference compound.  It may have been left displaced from a prior calibration of the seduce1c pulse.  Relevant frequencies are posted on a card on the operator's console.

Check the pulse program for which the pulse is intended for information about the pulse.  It may be shaped or square, and a nominal power and pulse time will be recommended.  Check the comments in the calib15n or calib13c pulse program to see which pulse and power setting are to be used for the 13C or 15N pulse.  For those who learned on the AMX 500, the Avance systems have some changes in the names of parameters.  Power attenuation levels associated with square pulses p1, p2, etc. are now known as PL1, PL2, ... (not HL1, HL2, DL1, DL2, DBL0,DBL1,  etc.).  [Note: the actual commands are in lower case.]  Attenuation levels are on a scale that can be either positive or negative, rather than starting at 0.   AMX power levels vary between 0 and + 90 dB; Avance power levels vary between -6 and +120 dB).  Nominal power levels recommended for the AMX 500 will NOT be correct for the Avance 500 (or any of the Avance spectrometers).  Parameters related to shaped pulses start with sp instead of tp.  Also note from the pulse program of the experiment into which the pulse will be inserted whether the time is to be adjusted or the power level.  For some kinds of pulses, the time must be exactly as specified and the power level should be adjusted to yield a null (meaning flat) antiphase spectrum.  Power levels can be adjusted to 0.01 attenuation units on the Avance systems.  If this is not noted in the pulse program, it is typical to leave the power level at the recommended nominal level and adjust only the pulse time.  However, if the time increases too much, it is possilbe that the pulse program will not accomidate it. So for an increase of greater than about 10%, it may be safer to adjust the power to keep the time closer to the nominal time.  If the time increases greatly (eg 2x), then this indicates a problem of some kind.  Sometimes the discrepency is just due to a typo or obsolete documentation.  However, if your values are high compared to others you've recently taken, the problem could be poor tuning, or some degradation in the performance of the equipment.  If the problem is not resolved by repeating tuning, etc., it should reported.  If adjusting power, remember that lower (including more negative) attenuation values mean more power is applied.  With more power, less time will be needed to achieve a comparable rotation (e.g.. 90 degrees).

First do the experiment with 0 pulse time.  For a shaped pulse, 0 is not a legal time, so use the estimated 180 instead.  Transform with ef.  In each case, the large signal you will observe is from the solvent.  Flanking the solvent peak will be a smaller antiphase doublet (one peak up and one down) that is 90 degrees out of phase with the solvent.  You may need to enlarge the scale to observe the doublet.  Phase the doublet and save the phase [save and return].  Subsequently, zoom on the doublet, click <dp1>, agree with the 3 popup messages, run paropt through the 180 degrees estimated from the nominal 90 degree time to crudely verify the position of the 90 degree pulse.  Then run paropt with finer intervals around the 90 degree point to get a closer determination.  If you scale up the signal at the null point, you will notice a residual out of phase signal probably originating from  an imprecision in the delay time creating the antiphase signal.  In the the 13C calibration, there is a 3rd out of phase signal between these two originating from incomplete 13C labeling.  Andy prefers to judge the null by the residual signals appearing symmetric with equal intensity above and below the baseline.  We don't have a good accounting of the tolerance with which the pulse times are required for the different experiments, although about 0.5% - 1% might be a good rule of thumb in the absence of more specific information.  If you want to try to set the times more closely, I recommend reobtaining the data at the best estimate of the null, and phasing on the residual signals, then save phase. For 13C, use the middle peak, which will be antiphase to the other two.  For 15N, make the pattern symmetrical.  Then take  readings through the null reading by zg and efp.  Unless the signal you are trying to null is completely gone, there will be a small dip in the baseline either on the inside edges or the outside edges of this phased residual signal.  I notice a disturbance (splitting) in one of the 13C peaks, and use the other.  Also, trying to calibrate precisely by this method may require more precise shimming if there are shape perturbances around the base of the peak.

draft: 6/23/02 - Steve Hardies