Back to Analysis of Direct and Reverse Titrations


Tight, slow-off binding coupled with isomerization in intermediate exchange

Contents

 

Here we will compare outcomes of the four alternative models for conditions when binding is very slow while isomerization is in intermediate exchange. I choose binding affinity of 108 M-1 and off-rate of 1 s-1. R and RL are separated by 100 s-1 and R* or R*L - by 100 s-1 from corresponding species. Isomerization equilibrium is shifted 5:1 toward *-forms and occurs in intermediate exchange with kex equal to the the chemical shift separation of the alternative conformations: 100 s-1   and the corresponding reverse rate: 100*1/6=17 s-1.

Location: As_Bi/

U_As.txt U_R_As_Bi.txt U_L_As_Bi.txt U_RL_As_Bi.txt

Exchange between alternative conformations is intermediate, so we don't see the peak of minor conformation any more. The system looks quite unpromising in terms of our ability to detect coupled conformational exchange. However, we may use fitting of this data with the U model as a research tool and observe effects on apparent Kd and koff .

 

Hypothesis: The key phenomenon we may exploit in analysis of this slow/intermediate-exchange system is the dependence of the fitted apparent Kd on the chemical shift separation between the peaks.

 

 

Back to Contents


The control fitting run: 2-site data fit using 2-site model

First we need to test the performance of the LineShapeKin 3 fitting algorithm using simulated data with the variable peak separation. We need to see how much the fitting quality is dependent on the separation between resonances.

As_chem_shift_dep/

50 s-1
100 s-1
200 s-1
300 s-1
600 s-1 900 s-1
  U_As_1 U_As_2 U_As_3 U_As_4 Simulate setup_wide U_As_5 Simulate setup_wide U_As_6
 
 
  Optimum norm: 2.96e-04
[1] Kd: 1.42672e-08 +/- 3.7e-09
[2] Koff: 9.73869e-01 +/- 9.8e-03
[3] Scale Factor: 2.04270e+00 +/- 8.9e-04
Optimum norm: 3.30e-04
[1] Kd: 1.41276e-08 +/- 2.2e-09
[2] Koff: 9.28704e-01 +/- 1.1e-02
[3] Scale Factor: 2.04250e+00 +/- 9.8e-04
Optimum norm: 1.07e-03
[1] Kd: 1.63449e-08 +/- 3.6e-09
[2] Koff: 9.06829e-01 +/- 2.1e-02
[3] Scale Factor: 2.02114e+00 +/- 1.9e-03
Optimum norm: 7.02e-03
[1] Kd: 2.97818e-08 +/- 1.1e-08
[2] Koff: 9.73446e-01 +/- 5.1e-02
[3] Scale Factor: 2.05523e+00 +/- 5.1e-03
Optimum norm: 1.30e-02
[1] Kd: 1.46075e-06 +/- 4.7e-07
[2] Koff: 8.65382e-01 +/- 6.1e-02
[3] Scale Factor: 1.89143e+00 +/- 8.5e-03
LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 2.50e-02
[1] Kd: 7.40478e-06 +/- 1.4e-06
[2] Koff: 9.98975e-01 +/- 9.2e-02
[3] Scale Factor: 1.83413e+00 +/- 1.2e-02

I collect these numbers in summary_Kdvsw.txt and plot with

plotdatasets 'Fit_quality' 'Chemical shift separation, s^{-1}' 'K_d, M'

A note on fitting quality: Fitting result is quite stable to about 300 s-1. Pre-set Kd was 1*10-8. After that the fit quality degrades and that is manifested by deviation of the fitted Kd from the preset Kd value.

 

Back to Contents

 

 


U_R model: Fitting of the data simulated with different chemical shift separations

Common parameters we use are:

# Association constants
Ka_names A B
Ka 1e8 5

# Rate constants of REVERSE reactions
k_names A B
k2 1 25


             
between peaks/dw(B) 50/25 100/50 200/100 300/150 300/50 50/300
 

Simulate setup U_R_As_Bi_1

Simulate setup U_R_As_Bi_2

Simulate setup U_R_As_Bi_3

Simulate setup U_R_As_Bi_4

Simulate setup U_R_As_Bi_5

Simulate setup U_R_As_Bi_6
 
 
  LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 6.07e-04
[1] Kd: 2.83848e-08 +/- 9.4e-09
[2] Koff: 7.56248e-01 +/- 1.7e-02
[3] Scale Factor: 2.00888e+00 +/- 1.5e-03
Maximum number of iterations reached!
Determination of parameter uncertainties...
Norm (sum of squared residuals): 1.18e-03
[1] Kd: 1.58792e-08 +/- 5.3e-09
[2] Koff: 6.97622e-01 +/- 2.0e-02
[3] Scale Factor: 2.04771e+00 +/- 1.9e-03
LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 3.23e-03
[1] Kd: 2.28037e-06 +/- 3.6e-07
[2] Koff: 4.80843e-01 +/- 3.1e-02
[3] Scale Factor: 2.08739e+00 +/- 5.5e-03
LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 5.65e-03
[1] Kd: 1.51640e-05 +/- 1.2e-06
[2] Koff: 5.22583e-01 +/- 4.4e-02
[3] Scale Factor: 2.20645e+00 +/- 8.8e-03
LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 6.04e-03
[1] Kd: 3.54197e-06 +/- 5.2e-07
[2] Koff: 6.64121e-01 +/- 4.2e-02
[3] Scale Factor: 2.11448e+00 +/- 6.9e-03
LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 1.55e-03
[1] Kd: 6.35537e-08 +/- 2.0e-08
[2] Koff: 8.99030e-01 +/- 3.6e-02
[3] Scale Factor: 2.11083e+00 +/- 2.8e-03

Summarize Kd vs. dw in summary_Kdvsw.txt

plotdatasets 'Fit_quality' 'Chemical shift separation, s^{-1}' 'K_d, M'

move files to U_R/

We observe a very remarkable phenomenon. The line shapes at the larger chemical shift separations resemble 2-site exchange less so the fit quality becomes progressively poorer. The apparent dissociation constant increases with dw very rapidly by 2-3-orders of magnitude at 300 s-1.

Important that it is chemical shift separation between R and RL what matters, not between R and R*.

 

 

 

Back to Contents


 


U_L model: Fitting of the data simulated with different chemical shift separations

Test the same common parameters:

# Association constants
Ka_names A B
Ka 1e8 5

# Rate constants of REVERSE reactions
k_names A B
k2 1 25


between peaks/dw(B) 50 100 200 300
 

Simulate setup U_L_As_Bi_1

Simulate setup Simulate setup U_L_As_Bi_3 Simulate setup U_L_As_Bi_4
 
 
  LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 2.39e-04
[1] Kd: 6.61183e-08 +/- 7.7e-09
[2] Koff: 9.68117e-01 +/- 1.0e-02
[3] Scale Factor: 2.04040e+00 +/- 9.6e-04
LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 6.57e-04
[1] Kd: 8.10815e-08 +/- 9.9e-09
[2] Koff: 9.20072e-01 +/- 1.7e-02
[3] Scale Factor: 2.01947e+00 +/- 1.7e-03
LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 1.37e-03
[1] Kd: 7.72719e-08 +/- 1.1e-08
[2] Koff: 9.47301e-01 +/- 2.6e-02
[3] Scale Factor: 1.99084e+00 +/- 2.5e-03
LSQCURVFIT converged to a solution
Determination of parameter uncertainties...
Norm (sum of squared residuals): 3.08e-03
[1] Kd: 9.67200e-08 +/- 1.5e-08
[2] Koff: 9.77071e-01 +/- 3.9e-02
[3] Scale Factor: 1.98327e+00 +/- 3.9e-03

Summarize Kd vs. dw in summary_Kdvsw_U_L.txt

plotdatasets 'Fit_quality' 'Chemical shift separation, s^{-1}' 'K_d, M'

It is remarkable that the reverse titration of the U_R system produces very consistent Kd over a range of chemical shifts.

Practical conclusions

It may be predicted that if we can fit line shapes to obtain Kd for a number of sites in the ligand and the receptor, then due to intrinsic spread of chemical shifts among the sites, we may be able to propose that one of the binding partners undergoes conformational change in the free form if we see significant (orders of magnitude) spread in the Kd values of one of them while Kd values of the other are clustered together.

Such situation will be an indication that a more complex model must be tried to account for the data.

 

 

 

Back to Contents

 

 


U_RL model: Fitting of the data simulated with different chemical shift separations

 

 

between peaks/dw(B) 50/25 100/50 200/100 300/150 300/50 50/300
 

Simulate setup U_RL_As_Bi_1

Simulate setup U_RL_As_Bi_2

Simulate setup U_RL_As_Bi_3

Simulate setup U_RL_As_Bi_4

Simulate setup U_RL_As_Bi_5

Simulate setup
  Optimum norm: 1.34e-04
[1] Kd: 1.14250e-08 +/- 5.5e-09
[2] Koff: 1.38913e-01 +/- 6.4e-03
[3] Scale Factor: 2.01935e+00 +/- 7.4e-04
Optimum norm: 1.97e-04
[1] Kd: 1.24050e-07 +/- 2.3e-08
[2] Koff: 1.17300e-01 +/- 8.0e-03
[3] Scale Factor: 2.05128e+00 +/- 1.1e-03
Optimum norm: 1.05e-03
[1] Kd: 1.64103e-09 +/- 7.7e-09
[2] Koff: 2.36669e-01 +/- 2.3e-02
[3] Scale Factor: 2.10072e+00 +/- 2.3e-03
Optimum norm: 2.62e-03
[1] Kd: 5.13734e-09 +/- 2.2e-08
[2] Koff: 2.44385e-01 +/- 3.7e-02
[3] Scale Factor: 2.09472e+00 +/- 3.8e-03
Optimum norm: 1.30e-03
[1] Kd: 1.30760e-06 +/- 1.8e-07
[2] Koff: 2.32323e-01 +/- 2.3e-02
[3] Scale Factor: 2.04031e+00 +/- 3.3e-03
Optimum norm: 4.27e-03
[1] Kd: 8.29264e-09 +/- 4.5e-08
[2] Koff: 2.22045e-14 +/- 3.4e-02
[3] Scale Factor: 2.10051e+00 +/- 4.6e-03

Conclusion: Tight binding with isomerization of the complex does not leave much trace in the line shapes no matter what chemical shift difference between the end species.

 

 

 

Back to Contents


Conclusions

  1. Tight binding with slow off-rate presents very limited information for resolving the mechanisms.
  2. However, the feature that may be used is dependence of apparent 2-site Kd on the chemical shift difference between the peaks observed in U_R (direct titration) but not in U_L model (reverse titration). Thus dependence of Kd on dw is a sign of isomerization in a receptor.
  3. U_RL does not have any dependence of Kd on dw

 

 

Back to Contents

 

Back to Analysis of Direct and Reverse Titrations