Practical Manual

General outline to use the structural information obtained from molecular

alignment
1. In order to use this information one needs to know the direction and the size of the tensor (susceptibility,
alignment, etc).
2. Minimization of the deviation between the measured quantity and its calculated value from a given
structure and set of tensor parameters.
3. One has to be able to evaluate the outcome of the minimization.

General outline for the minimization procedure

1. Get a good estimate on the tensor parameters (anisotropy and rhombicity/asymmetry).
2. Define structural constraints with respect to the initially arbitrary tensor coordinate system.
3. Turn on the force constant for a particular alignment potential such that the final RMS between the
measured and calculated values reflects the experimental error.
4. During the calculation the tensor orientation will be automatically determined through global
minimization with respect to the current structure.
5. Refine the initial estimate of the tensor parameters. This can be achieved through either grid search
method or built into the minimization it self.

Dipolar coupling refinement using Xplor-NIH

The measured dipolar couplings depend on the relative orientation of the dipolar vectors with respect to
the alignment coordinate frame ( and ), the magnitude of alignment tensor (Aa, Ar; or Da, R) the
gyromagnetic ratios of the interacting nuclei (p and q), the distance between them (rpq), and the
generalized order parameter of the dipolar vector (S).

Dpq - S pq [Aa (3cos2 1) + 3/2 Ar (sin2 cos2)] / r3pq (1)

In practice the above equation is usually reduced to:

Dpq = Dapq [ (3cos2 1) + 3/2 R (sin2 cos2) ]. (2)

1
The Dapq will depend on the p and g as well as the distance between the two nuclei rpq. Therefore
when one compares different types of dipolar couplings, they need to be normalized according to their
Dapq values.

In Xplor-NIH the function that is being minimized is the difference between calculated and measured
dipolar couplings.

E = k (Dmeas Dcalc)2 (3)

The variable k is a force constant that can be adjusted during the calculation.

I A. Obtain a good estimate on the magnitude of the alignment tensor: Da and R

In order to get a good estimate for Da and R one would need to create a histogram that reflects
the distribution of the measured dipolar couplings. The extremes of the histogram correspond to the
alignment tensor components Axx, Ayy, and Azz ( |Azz| |Ayy| |Axx| ). Also note that the alignment tensor
is traceless, that is Axx + Ayy + Azz = 0. Da and R can be computed with the following equations:

Azz = 2 Da (4)

Ayy = - Da (1 + 3/2 R) (5)

Axx = - Da (1 3/2 R) (6)

The larger the number of dipolar couplings used in creating the histogram the better the estimate is
going to be. In order to use different types of dipolar coupling data (CH, NH, CC, NC, etc) to create
one histogram, one needs to first scale them by their corresponding and r values so that they are all
normalized with respect to one type of dipolar couplings. Residues with substantially lower order
parameter than average should be excluded from the histogram.

Exercise:
1. Go to the subdirectory dipolar_fit and find the files: NH.data and CH.data

2. Residues after L71 has order parameters that are lower than average and they must be

excluded from the histogram. Type: mstat NH.data 1 and take a look at the resulting

histogram. The histogram is rotated by 90. Do the same with the CH.data file and note the

extremes on these histograms. Note mstat is a program that is a component of nmrPipe

package.

3. Given that C/N = -2.4808, rNH=1.02, and rCH=1.1 using equation (1) estimate the scaling

2
 factor for DCH versus DNH.

4. Go ahead and scale the CH.data so that it is normalized with respect to NH dipolar

couplings.

Type scale.awk CH.data > CH_scaled.data. The new normalized values are in

the file CH_scaled.data.

5. Combine the NH and CH dipolar couplings into one file.

Type cat NH.data CH_scaled.data > all.data

6. Type mstat all.data 1 to get the final histogram.

7. Get the values for Axx, Ayy, and Azz from the histogram. Go ahead and calculate Da and R.

I B. Fitting dipolar coupling data to an existing structure

If the protein structure already exists or if you think there is a similar structure available you can go
ahead and fit your dipolar coupling data to the existing structure. This process is identical to fitting the
relaxation data to a protein structure to get the anisotropic diffusion parameters.

Exercise:
1. Go to the directory dipolar_fit and find the file ubq.pdb.

2. Calculate the directional cosines for all of the NH bonds in ubiquitin the same way as in the

relaxation analysis.

Type NH.awk ubq.pdb > ubq.cos

3. The direction of all of the NH bonds in the xray coordinate frame is now contained in the file

ubq.cos.

4. Open the file NH.data in a text editor. Delete residues after L71 since they have low order

parameter and their DNH will be scaled down.

5. Count the number of residues in NH.data file.

6. Fit the dipolar data:

3
 Type align_powell NH.data 51 ubq.cos. Use the following starting values: Da and R

found in the previous exercise, = 40, = 40, and = -30. The experimental error in

DNH is approximately 0.1 Hz.

8. Using the optimized alignment tensor obtained in step 6 back calculate the DNH using the

available structure.

Type align_powell_back NH.data 51 ubq.cos result.out. Use the optimized alignment

tensor parameters.

9. Inspect the difference between experimental and calculated dipolar couplings in the file

result.out. One can also plot them out using gnuplot:

Type gnuplot

Within gnuplot type: plot result.out using 2:3

To exit gnuplot type: quit.

4
II. Define axis representing the alignment tensor coordinate system

In order to calculate the dipolar coupling for a pair of nuclei from a given structure one needs to
be able to calculate the angles and . These angles are measured between the dipolar vector and the
alignment coordinate system. In Xplor-NIH the coordinate system is represented by four pseudo atoms:
OO, X, Y, and Z. OO represents the origin for the coordinate system, while X, Y, and Z represent the
Cartesian axis system. The projection of the dipolar vector onto the OO-Z axis results in cos, while
projection onto OO-X and OO-Y will result in sin cos and sin sin, respectively. Different
alignment tensor will have its own axis system.

Considerations in setting up the axis system:

1. It has to be orthogonal.
2. It has to be positioned far enough away from the molecule to be calculated such that non-
bonded interactions (steric, electrostatic, etc) are not present between them.
3. It has to be able to rotate freely without translation.

Exercise:

1. Go to the subdirectory dipolar_csa and find the file axis.psf.

2. In this file the axis system is defined. Take a look at axis.psf file.

3. The file par_axis_3.pro defines the geometry and energetic of the axis system.

4. Find the file axis.pdb. Open it with a text editor and examine the coordinates of the atoms as

well as the numbering of the atoms. This axis file has to be combined with the protein

coordinate. Find and open the file ubq.pdb and go to the end of the file and notice the axis

5
 coordinates.

5. Create the second axis system:

Type cp axis.pdb axis_501.pdb.

Edit axis_501.pdb and change the residue numbers from 500 to 501.

Also move the axis system by 100 along the z direction in the pdb frame, save the file and

exit from the editor. Edit the file ubq.pdb and copy and paste axis_501.pdb entries into the end

of ubq.pdb.

Edit the atom numbering to correspond to the new addition of axis system, save the new

ubq.pdb file and exit the text editor.

Type cp axis.psf axis_501.psf

Edit the axis_501.psf file. Look for the line containing !NATOM. Changed the number 500 to

501 in the next following lines, save the file and exit the editor.

6. If possible open the ubq.pdb file with a molecular graphic program and take a look at where the

axis are relative to the protein.

6
III. Creating dipolar constraint file

The constraint file for dipolar coupling in XPLOR-NIH should contain the axis system representing the
alignment tensor, the dipolar interacting nuclei, measured dipolar coupling, and experimental error of
the dipolar coupling. An example of a constraint entry:

assign ( resid 500 and name OO )

( resid 500 and name Z )
( resid 500 and name X )
( resid 500 and name Y )
( resid 2 and name HN )
( resid 2 and name N ) -8.182 0.6 0.5

This constraint is to the alignment tensor associated with axis 500 and for residue 2. The measured NH
dipolar coupling is 8.182 and the estimated experimental errors are 0.6 and + 0.5. The dipolar
couplings of mobile residues have to be treated differently. The use of square open potential would be
more appropriate in this case. This can be achieved by extending the positive error to very large value
when the measured dipolar coupling is positive, and by extending the negative error when the dipolar
coupling is negative.

Exercise:

1. Go to the subdirectory dipolar_csa/tables.

2. Used the first three entries in CH.data file to create a constraint table with axis system 500 and

experimental error of 0.2 Hz.

3. Compare your file to the CH.tbl file. Note that the order for the atoms OO, Z, X, and Y are

important, while order for the dipolar interacting atoms are not.

4. Create NH dipolar constraints for the following mobile residues (S2 < 0.5):

R74 5.459 Hz

G75 2.114 Hz

G76 1.137 Hz.

The estimated experimental error was 0.5Hz.

5. Compare your table to nh_mob.tbl file.

6. Create CH dipolar constraints for the following residues:

7
 Q2 4.727 Hz

I3 8.186 Hz

F4 16.414 Hz.

These were measured in a different alignment medium and the experimental errors were 0.5

Hz. Use the axis system created in Part II step 5.

7. Compare your constraints to CH2.tbl file.

8
IV. Setting up XPLOR-NIH input file for dipolar coupling refinement

At this point you should already have the estimate for Da and R; constraint tables, and the axis
system(s) set up. In the input file you have to define the potential to be used for refinement, restraining
the axis system from translation, force constant to be used during the simulated annealing protocol.

Exercise:

1. Go to the subdirectory dipolar_csa and find the file refine.inp.

2. Locate where the axis.psf, par_axis_3.pro, and ubq.pdb files are read in.

3. Locate where Da and R are defined. Check with the values that you estimated in Part I step 6.

4. Locate where the dipolar potentials are being setup. Check for the potential type, force constant

variable, and averaging being used in the refinement. What are the differences between the

dipolar classes jch and jch2. What are the similarities and differences between dipolar

classes side and jnh2.

5. Locate the line containing constraints fix statement. Identify which atoms are being fixed and

explain why?

6. Locate the variables: ini_sani, fin-sani, and sani_fac they define the force constant initial,

final, and how fast it is being ramped up during the calculation. Find out the assigned values for

these variables.

7. Find out how these variables are related to the dipolar force constants (ksani).

9
V. Running dipolar coupling refinement with XPLOR-NIH.

At this point you should be ready to run the refinement.

1. Type xplor < refine.inp > refine.log.

2. Check the log file as the calculation is running to make sure that there is no error during the

refinement.

3. Once the calculation is finished, check the log file. Towards the end the agreement between

measured and calculated dipolar couplings will be printed out.

4. Make sure that the RMSD is close to the estimated experimental error.

5. Check the gradient energy during the calculation by typing grep grad(E) refine.log >

grad.out.

6. Open the file grad.out with a text editor and examine the gradient energy as the calculation

progressed.

7. Check the resulting structure ubq_dipo1.pdb using a molecular graphics package.

8. Make sure that the axis systems are not distorted.

9. Make sure that other potential energies are not increased substantially.

10
Hydrogen Bond Angle and Distance Correlation refinement using Xplor-
NIH
We found a correlation between distance and angle of hydrogen bonds in protein through a sampling of
high-resolution Xray structures (resolution < 1.0 ). This was confirmed through ab initio calculation.
An empirical potential was derived to limit sampling of hydrogen bond geometry during an xplor
calculation to increase efficiency and more accurate representation of hydrogen bond in protein. The
empirical formula is:

1/ R3 = A + [B / {2.07 + cos NHO}3]

where A equals 0.019 and B is 0.21 3. In Xplor-NIH the target function for hydrogen bond angle and
distance correlation is:

E = k ( 1/ R3 - A - [B / {2.07 + cos NHO}3] )2

when the term inside the parenthesis is larger than zero, otherwise E is set to zero. This drives the
correlation between the hydrogen bond distance and angle to be under the correlation curve.

The use of hydrogen bond angle and distance correlation in XPLOR-NIH is quite straightforward. The
hydrogen bond is defined in the constraint table as:

assign ( resid 4 and name N )

( resid 4 and name HN)
( resid 65 and name O ).

During the xplor refinement a force constant associated with the hydrogen bond potential is increased
gradually during the calculation.

Exercise:
1. Go to the subdirectory dipolar_csa.

2. Identify some hydrogen bonds in ubq.pdb file using a molecular graphics package.

3. Create a constraint table reflecting those hydrogen bonds found. No bifurcated hydrogen bonds

should be included at this time. Compare your table to the file hbda.tbl in the dipolar_csa/tables

directory.

4. Open the refine_hbda.inp file using a text editor.

5. Locate where the hbda potential is defined.

6. Locate the variables ini_hbda, fin_hbda, and hbda_fac and find their values.

7. Notice again how they are related to the force constant used in the refinement (khbda).

11
8. Run xplor calculation by typing: xplor < refine_hbda.inp > refine_hbda.log

9. When the calculation is done, check the log file for any errors.

10. Check the log file for the RMSD of the HBDA term.

11. Make sure that no other potential energies are disturbed.

12. Finally check for the gradient energy of the refinement as in dipolar coupling refinement part V

step 5 and 6.

13. Examine the resulting structure using a molecular graphics package.

12
 Chemical shift anisotropy refinement using Xplor-NIH
In the event that a small degree of alignment can be introduced to the molecule under study any second
rank tensor of the Hamiltonian will not be averaged to zero. For instance the dipolar coupling is no
longer averaged to zero and results in small but measurable residual dipolar coupling. Similarly
chemical shift anisotropy tensor will also not be averaged to zero. In this case a small difference in
chemical shift can be observed between the isotropic and anisotropic samples. The difference
corresponds to the projection of the CSA tensor onto the alignment tensor, and it can be expressed as:

!" = $ $A jj cos2 # ij"ii

i= x , y,z j = x, y, z

where ij is the angle between Ajj principal axis of the alignment tensor and the ii principal axis of the
CSA tensor. This residual CSA contains useful structural information that can easily be used in a
structure refinement.

In Xplor-NIH the target function for the CSA refinement is:

E = k (meas calc)2

where k is the adjustable force constant, while meas and calc are the measured and calculated ,
respectively.

I. Creating CSA constraint file

The constraint file follows the same convention as in the dipolar coupling case. The axis system
representing the alignment tensor has to be set up, and Da and R have to be estimated. The format of
the constraint for carbonyl CSA is as follows:

Assign ( resid 500 and name OO )

( resid 500 and name Z )
( resid 500 and name X )
( resid 500 and name Y )
( resid 2 and name C )
( resid 2 and name O )
( resid 3 and name N ) -38.36 5.00 5.00

in this case the atoms C, O, and N defines the local CSA coordinate system. This particular entry for
carbonyl CSA of residue 2 with respect to alignment tensor represented by residue 500. The local CSA
coordinate system is given below.

13
 In contrast to dipolar coupling constraint that deals with a vector connecting the two dipolar interacting
nuclei, the CSA value is a function of tensor projection, thus it has to be defined by the three atoms to
reference the local CSA geometry. The format for nitrogen CSA is:

assign ( resid 500 and name OO )

( resid 500 and name Z )
( resid 500 and name X )
( resid 500 and name Y )
( resid 2 and name C )
( resid 3 and name N )
( resid 3 and name HN ) -73.28 5.00 5.00

that restraints nitrogen CSA of residue 3 due to alignment represented by residue 500. The local CSA
coordinate system is given below:

14
Exercise:

1. Go to the subdirectory csa.

2. Create constraints for the first three entries in the ncsa.data file. This file contains nitrogen

CSA values. Use residue 500 to represent the axis system. Use 5 ppb as the experimental error.

3. Create constraints for the first three entries in the ccsa.data file. Use the same axis system as

the nitrogen csa. Use 5 ppb as the experimental error.

4. Go to the subdirectory dipolar_csa/tables and compare your files with ncsa.tbl and ccsa.tbl

files.

II. Setting up XPLOR-NIH input file for CSA refinement

The steps needed here are the same as in the dipolar coupling refinement. The axis system(s), constraint
files, as well as estimated Da and R should already be available. In the input file one has to set up the
potentials and force constants for various CSA restraints to be used in the refinement as well as
ensuring that there is no translation of the axis system. The only difference between the CSA and
dipolar coupling set up in XPLOR-NIH is that one needs to define the chemical shift tensor components
(11, 22, and 33).

Exercise:
1. Go to the subdirectory dipolar_csa.

2. Find and open refine_csa.inp file in a text editor.

3. Locate the section in which the CSA potential is defined (DCSA).

4. Evaluate the differences between the nitrogen and carbonyl CSA setup.

5. Find out how the force constants are being ramped up (kcsa).

6. At this point you should be able to start the CSA refinement by typing xplor < refine_csa.inp

> refine_csa.log.

7. When the calculation has finished, take a look at the refine_csa.log file and check for energies

as well as final RMSD for the CSA restraints. Also check for the energy gradient (see dipolar

15
 coupling refinement part V steps 5 & 6).

8. Take a look at the resulting structure (ubq_csa1.pdb) using a molecular graphics package and

make sure that the axis and the molecule are not distorted.

16
 Diffusion Anisotropy refinement using Xplor-NIH
A non-spherical molecule will have an anisotropic rotational diffusion tensor. The dipolar relaxation
rates of two interacting nuclei will depend on how the vector connecting them is oriented relative to the
diffusion tensor. This orientation dependence can structurally be useful. In practice one needs an
anisotropy in rotational diffusion that is substantially large to make it useful. This is limited by the error
in the measurement. As a rule of thumb, one would need at least one order of magnitude between the
measured quantity to be used in refinement and its experimental error.

Dipolar relaxation rates for two spins 1/2 (such as N-H) are given as:

1/T1 = d2 [ J(H - N) + 3 J(N) + 6 J(H + N) ] + c2 J(N)

1/T2 = 0.5 d2 [4 J(0) + J(H - N) + 3 J(N) + 6 J(H) + 6 J(H + N)] + (1/6) c2 [ 3

J(N) + 4 J(0)]

NOE = 1 + ( H /N ) d2 (6 J(H + N) - J(H - N)] T1

where

d2 = 0.1 [(H N h) / (2 r3NH)]2

c2 = (2/15) [N 2 ( //- )2]

Spectral density is given by the fourier transform of the correlation function.

J() = - C(t) e-it dt

For an isotropic diffusion:

J() = [ Sf2 c / (1+(c)2) + (1-Sf2) / (1+(')2 ]

1/ = 1/ c + 1/ e

For an anisotropic diffusion (axial symmetry):

J() = S2 ( A1 1 / [1 + (1)2] + A2 2 / [1 + (2)2] + A3 3 / [1 + (3)2] ) + (1-S2) / [1 + ( )2]

A1 = 0.75 sin4 1/ 1 = ( 4 D// + 5 D )

A2 = 3 sin2 cos2 1/ 2 = ( 5 D// + D )
A3 = (1.5 cos2 - 1) 1/ 3 = ( 6 D )

where is the angle between the dipolar vector and the principal axis of diffusion tensor. To a first
approximation the ratio of T1 and T2 is not sensitive to fast motion (S2 and ) and variation in CSA
values. From the above equation variations in T1/T2 will be mostly due to the angle . When the
rotational diffusion tensor is fully asymmetric the rates will depend on two angles describing the

17
orientation of the dipolar vector in the diffusion tensor frame and three diffusion rates. Xplor-NIH treats
the diffusion anisotropy as fully asymmetric. You have to set asymmetry parameter () to one to reduce
the diffusion tensor to an axially symmetric one.

The target function for diffusion anisotropy refinement in Xplor-NIH is:

E = k [(T1/T2)meas (T1/T2)calc ]2

where k is the adjustable force constant.

I. Creating diffusion anisotropy constraint file

The constraint file follows the same convention as in the dipolar coupling case. While in the dipolar
coupling case one refines an orientation of a vector in an alignment tensor frame, in the diffusion
anisotropy case it is a refinement of a vector in the diffusion tensor frame. One would need an axis
system. In this case the axis will represent the diffusion tensor. The format of the constraint file is
exactly the same as in the dipolar coupling case, except the last two atoms assigned belong to the
dipolar relaxing ones,

Assign ( resid 502 and name OO )

( resid 502 and name Z )
( resid 502 and name X )
( resid 502 and name Y )
( resid 3 and name N )
( resid 3 and name HN ) 6.8860 0.2500

the error in the constraint file is taken as error. It is important to make sure that residues experiencing
large amplitude fast motion or exchange be eliminated from the restraint table. These residues will
typically be over estimating the T1/T2 ratio.

Exercise:

1. Go to the subdirectory dipolar_csa/tables and take a look at the file t1_t2.data. It contains the

T1/T2 ratios for LARGE ubiquitin acquired on a 600 MHz NMR spectrometer.

2. The relaxation rates do depend on second order polynomial of the directional cosines (cos),

thus one would expect to see similar type of histograms as in the dipolar case (dipolar, part I

step 2). Try to create the histogram of the T1/T2 ratio and describe the diffusion tensor.

3. From the histogram determine the minimum, maximum, and average T1/T2 values.

4. Run the program diff_estimate to try to estimate the diffusion parameters: c, D// / D, and

18
 (asymmetry).

5. Find and open the file diff_anis.tbl with a text editor and examine the restraint table. Note the

residue used to represent the diffusion tensor axis system.

6. Go ahead and create a new axis system represented by residue 502. See dipolar coupling

refinement part II step 5.

II. Setting up XPLOR-NIH input file for diffusion anisotropy refinement

At this point you should have everything that you would need to create an input file for Xplor-NIH. The
rest is the same routine that you have gone through to set up the input file for dipolar coupling as well
as the CSA refinement.

Exercise:

1. Go to the subdirectory dipolar_csa.

2. Find and open the input file refine_dani.inp in a text editor.

3. Locate the section where the diffusion anisotropy potential is defined (DANI).

4. Examine the coefficients used and compare to what you found in part I.

5. Check the force constant being used and make sure that they are being incremented properly

(kdani).

6. At this point you can start the refinement by typing xplor < refine_dani.inp >

refine_dani.log.

7. If possible check for the energy gradient, other potential energies, as well as the resulting

structure ubq_dani1.pdb for any irregularities.

19