Measurement and modeling of the acid dissociation constants of ...

Electronic Supplementary Material (ESI) for Dalton Transactions This journal is © The Royal Society of Chemistry 2013

Measurement and modeling of the acid dissociation constants of tripeptides containing Glu, Gly, and His using potentiometry and generalized multiplicative analysis of variance Rima Raffoul Khoury1, Gordon J. Sutton1, D. Brynn Hibbert1*, Diako Ebrahimi2* 1

School of Chemistry, The University of New South Wales, Sydney NSW 2052, Australia

2 Centre for Vascular Research, Faculty of Medicine, The University of New South Wales, Sydney NSW 2052, Australia Supporting information

Bibliography of pKa values for tripeptides considered in this paper This bibliography was kindly provided by Professor Kipton Powell from an IUPAC archive1. cpeptide/M 0.20 0.00 2.25

log 1 log 2 log 3 7.93 11.25 8.09 11.31 8.30 11.69

log 4

log 5

Peptide GGG GGG GGG

Medium KCl NaNO3

t /°C 25 25 25

Ref

GGH GGH GGH

KNO3 KCl KCl

25 25 25

0.10 0.20 0.16

7.93 7.96 8.22

14.64 14.60 15.09

17.90 17.52 17.93

5

GHG GHG

KCl KCl

25 25

0.10 0.20

8.02 7.98

14.55 14.33

17.67 17.05

8

2 3 4

6 7

6


9

GHG GHG

NaClO4 KNO3

25 37

3.00 0.15

8.62 7.68

16.16 14.04

19.73 17.07

10

HGG HGG

KNO3

25 21

0.10 0.00

7.34 7.62

12.73 13.14

15.83 16.31

11

EEE

KCl

25

0.20

8.02

13.00

17.29

5

21.14

24.18

12

Structures of the 27 tripeptides Structures of the tripeptides are given in the excel file provided in alphabetic order grouped in E series, G series and H series according to the type of the amino acid in terminal amine position. The peptide structures were drawn using ChemBioDraw Ultra 12.0 Suite Software

Titration curves for the 27 tripeptides A titration curve for each tripeptide is shown in the excel file provided in alphabetic order grouped in E series, G series and H series according to the type of the amino acid in the terminal amine position.

GEMANOVA models output for pKa for the terminal carboxylic acid proton The GEMANOVA outcome of all the models tested showing the model terms, the root mean square error of cross validation RMSECV and the number of estimated parameters of each model are shown in Table S1. A model term is coded with f1, f2, f3 representing parameters for the amino acids at the left hand end (f1), the middle (f2), and the right hand end (f3) of the tripeptide. The symbol ‘x’ indicates that the position has values for each amino acid to contribute to the pKa, and the symbol ‘1’ indicates a constant term. Note that each position (x) contributes three parameters to the model, and a complete constant term () requires one parameter. The model discussed in the text of the paper is highlighted in the table. Values of RMSECV > 1 indicate the model did not converge.


Table S1: Outcome of leave-one-out cross validation of all models of the pKa of the right-hand carboxyl proton with up to three terms. The model is described with terms with a parameter for each amino acid (E, G, H) in each position, where ‘1’= parameters are held constant; ‘x’ = a parameter is fitted for each amino acid.

RMSECV 0.30164 0.3658 0.35451 0.39081 0.51673 0.51458 0.51758 0.51338 0.48015 0.58689 0.51734 0.59708 0.39028 0.3074 0.37076 0.32454 0.38724 0.35167 0.38952 0.34022 0.37111 0.32641 0.37622 0.38282 0.39517 0.28716 0.30626 0.36586 0.37935 0.39081 0.33293 0.35885 0.36395 0.39081 0.61498 0.57748 0.57121 0.54399

Number of parameters 9 6 6 3 6 3 3 1 18 15 15 12 15 12 12 10 12 12 9 12 9 9 7 12 9 12 9 9 7 6 9 6 6 4 12 9 9 7


Number of RMSECV parameters 0.51458 6 0.51557 6 0.51458 4 0.51758 6 0.51758 4 9.732 27 0.61343 18 0.4021 18 0.51758 9 0.41104 18 0.51458 9 0.39081 9 0.51338 3 49.746 24 2.4609 24 86554 21 4.4387 24 0.93465 21 5.6735 21 4284.9 19 57806 21 0.41954 18 0.58289 15 0.50852 18 0.62553 15 0.35662 15 5.3157 13 0.85119 21 0.34953 18 0.40071 15 0.35297 18 0.32042 15 0.39927 15 0.39747 13 93717 15 0.52429 12 0.48684 12 0.33133 12 0.51557 9 0.36395 9 0.51758 7 8115 21 0.44624 18


Number of RMSECV parameters 0.38814 18 0.34684 15 0.41097 15 0.41286 15 0.41219 13 64236 15 0.54268 12 0.29425 12 0.51557 9 0.45941 12 0.35885 9 0.51458 7 15702 15 0.33814 12 0.40004 12 0.36395 9 0.37374 12 0.35885 9 0.39081 7 69131 11 0.53107 8 0.37065 8 0.51758 5 0.37396 8 0.51458 5 0.39081 5 1.335 21 1.36E+05 18 0.8554 21 1.25E+07 18 0.65229 18 0.40799 16 0.43121 15 0.34284 18 0.28641 15 0.32741 15 0.31726 13 0.34889 15 0.33306 12 0.38701 12 0.35461 10 0.32281 12 0.34485 12


Number of RMSECV parameters 0.32888 10 0.38863 12 0.36873 10 0.35885 7

Randomized control As a control, the experimental pKa values were assigned to randomly chosen tripeptides and then modeled. The best fit was, as expected, very poor which supports the argument that GEMANOVA modeling does capture information about the chemical patterns in this tri-peptide system. A graph of fitted against the randomly-assigned measured pKa of the control study is shown in Figure S1. The lack of fit indicates the fitted models described in the paper have captured information about the acidity of the proton from the distribution of amino acids.

pKa fitted by GEMANOVA

4.5 4 3.5 3 2.5 2 1.5 1.5

2

2.5

3

3.5

4

measured pKa

Figure S1. Randomized control. Best fit model of pKa values of the terminal carboxylic acid proton assigned to random tripeptides. Error bars are ± 2 standard deviations of the measured values.

GEMANOVA model for pKa of the terminal ammonium proton The optimum model for the pKa values for the left-hand terminal ammonium proton (pKa1) was pKa1 =  + aibj + ijk (i = E, G, H, j = E, G, H)

(S1)

where ai (i = ‘E’, ‘G’ or ‘H’) is the parameter for the left-hand residue with terminal ammonium, and bj (j = ‘E’, ‘G’ or ‘H’) is the parameter for the middle residue.


1.2

400

1

350

0.8

E G H

300

Coefficient

0.6

250

0.4

0.2

200

E G H

0

150

-0.2 100

-0.4

50

-0.6 -0.8 F1

F2 /1000

F3

0 F1

F2

F3

Figure S2. Parameters of the GEMANOVA model given in Eq. S1 for pKa of the left-hand terminal ammonium proton (pKa1).

pKa from GEMANOVA model 111-001

11 10.5 10 9.5 9

8.5 8 7.5 7 7

8

9 measured pKa

10

11

Figure S3. Plot of pKa of the left-hand terminal ammonium proton predicted by GEMANOVA model given in Eq. S1 against measured pKa. Error bars on the measured pKa are ± 2 × u(pKa) shown in Table 1 of text.


References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

L. D. Pettit and K. J. Powell, Stability Constants Database (SC-Database), (2010) IUPAC, Academic Software, Otley, U.K. C. G. Agoston, Z. Miskolczy, Z. Nagy and I. Sovago, Polyhedron, 2003, 22, 2607-2615. R. N. Goldberg, N. Kishore and R. M. Lennen, Journal of Physical and Chemical Reference Data, 2002, 31, 231-370. J. C. Pessoa, S. M. Luz, R. Duarte, J. J. G. Moura and R. D. Gillard, Polyhedron, 1993, 12, 2857-2867. E. Garribba, G. Micera, E. Lodyga-Chruscinska, D. Sauna and G. Sanna, European Journal of Inorganic Chemistry, 2005, 4953-4963. E. Farkas, I. Sovago, T. Kiss and A. Gergely, Journal of the Chemical Society-Dalton Transactions, 1984, 611-614. G. F. Bryce, R. W. Roeske and F. R. N. Gurd, Journal of Biological Chemistry, 1966, 241, 1072-1080. K. Takehara and Y. Ide, Inorganica Chimica Acta, 1991, 183, 195-202. R. Österberg and B. Sjöberg, Journal of Inorganic and Nuclear Chemistry, 1975, 37, 815-827. R. P. Agarwal and D. D. Perrin, J. Chem. Soc., Dalton Trans., 1975, 268-272. H. Aiba, A. Yokoyama and H. Tanaka, Bulletin of the Chemical Society of Japan, 1974, 47, 136-142. C. Kállay, K. Várnagy, G. Micera, D. Sanna and I. Sóvágó, Journal of Inorganic Biochemistry, 2005, 99, 1514-1525.