Artificial neural network modeling of peptide mobility and peptide mapping in capillary zone electrophoresis

Recently, we have developed an artificial neural network model, which was able to predict accurately the electrophoretic mobilities of relatively small peptides. To examine the robustness of this methodology, a 3-3-1 back-propagation artificial neural network (BP-ANN) model was developed using the same inputs as the previous model, which were the Offord's charge over mass term (Q/M(2/3)), corrected steric substituent constant (E(s,c)) and molar refractivity (MR). The data set consisted of 102 peptides with a larger range of size than that of our earlier report - up to 42 amino acid residues as compared to 13 amino acids in the initial study - that also included highly charged and hydrophobic peptides. The entire data set was obtained from the published result by Janini and co-workers. The results of this model are compared with those obtained using multiple linear regressions (MLR) model developed in this work and the multi-variable model released by Janini et al. Better predictive ability of the BP-ANN model over the MLR indicates the non-linear characteristics of the electrophoretic mobility of peptides. The present model exhibits better robustness than the MLR models in predicting CZE mobilities of a diverse data set at different experimental conditions. To explore the utility of the ANN model in simulation of the CZE peptide maps, the profiles for the endoproteinase digests of melittin, glucagon and horse cytochrome C is studied in the present work.     

Peptide mapping by capillary zone electrophoresis: how close is theoretical simulation to experimental determination

A multi-variable computer model is presented for the prediction of the electrophoretic mobilities of peptides at pH 2.5 from known physico-chemical constants of their amino acid residues. The model is empirical and does not claim any theoretical dependencies; however, the results suggest that, at least at this pH, peptides may be theoretically represented as classical polymers of freely joined amino acid residues of unequal sizes. The model assumes that the electrophoretic mobility can be represented by a product of three functions that return the contributions of peptide charge, length and width, respectively to the mobility. The model relies on accurate experimental determination of the electrophoretic mobilities of a diverse set of peptides, by capillary zone electrophoresis (CZE), at 22 degrees C, with a 50 mM phosphate buffer, at pH 2.5. The electrophoretic mobilities of a basis set of 102 peptides that varied in charge from 0.65 to 16 and in size from two to 42 amino acid residues were accurately measured at these fixed experimental conditions using a stable 10% linear polyacrylamide-coated column. Data from this basis set was used to derive the peptide charge, length, and width functions respectively. The main purpose of this endeavor is to use the model for the prediction of peptide mobilities at pH 2.5, and for simulation of CZE peptide maps of protein digests. Excellent agreement was obtained between predicted and experimental electrophoretic mobilities for all categories of peptides, including the highly charged and the hydrophobic. To illustrate the utility of this model in protein studies it was used to simulate theoretical peptide maps of the digests of glucagon and horse cytochrome c. The resulting maps were compared and contrasted with their experimental counterparts. The potential of this approach and its limitations are discussed.     

Prediction of quantitative calibration factors of some organic compounds in gas chromatography

A quantitative structure-property relationship (QSPR) methodology that involves multilinear (Hansch-type) and nonlinear (radial basis function neural network (RBFNN)) approaches was performed to correlate the quantitative molar calibration factors (f(M)) of 140 organic compounds against structural factors. The statistical characteristics provided by the multiple linear model (R(2) = 0.963; RMS = 0.089; AARD = 3.86% for test set) indicated satisfactory stability and predictive ability, while the predictive ability of the RBFNN model is somewhat superior (R(2) = 0.983; RMS = 0.075; AARD = 3.19% for test set). The multilinear model provided some insight into the main structure factors that modulate the quantitative calibration factor of the investigated compounds.     

Peptide mobility and peptide mapping in capillary zone electrophoresis. Experimental determination and theoretical simulation.

The electrophoretic mobilities of 58 peptides that varied in size from 2 to 39 amino acids and varied in charge from 0.65 to 7.82 are presented. The measurements were conducted at 22 degrees C using a 10% linear polyacrylamide-coated column and a 50 mM phosphate buffer at pH 2.5. Excellent separation of peptides and highly reliable peptide maps of protein digests are routinely obtained using these experimental conditions. The electrophoretic data were used to test existing theoretical models that correlate electrophoretic mobility with physical parameters. The results indicate that the Offord model that correlates electrophoretic mobility with the charge-to-size parameter q/M2/3 offers the best fit of our reliable experimental data. Furthermore, we also obtained the capillary zone electrophoretic profile of the endoproteinase Lys-C digests of a peptide sequencing standard, melittin, and horse myoglobin under the same experimental conditions as described above. The resulting peptide maps were compared with corresponding theoretical simulation.     

Modeling the electrophoretic mobility and diffusion of weakly charged peptides

A bead model to determine the electrophoretic mobilities and translational diffusion constants of weakly charged peptides is developed that is based on a approximate structural model of peptides and is also grounded in electrohydrodynamic theory. A peptide made up of X amino acids is modeled as N=2X beads with 2 beads representing each amino acid in the chain. For the two beads representing a particular amino acid in a peptide, the radius of one bead is set to one-half the nearest neighbor Calpha-Calpha distance, and the radius of the other bead is chosen on the basis of the diffusion constant of the free amino acid. Peptide conformations, which are defined by a set of psi-phi dihedral angles, are randomly generated by using the transformation matrix approach of Flory (Flory, P. Statistical Mechanics of Chain Molecules; John Wiley: New York, 1969) and rejecting conformations which result in bead overlap. The mobility and diffusion constants are computed for each conformation and at least 100 independent conformations are examined for each peptide. In general, the mobility is found to depend only weakly on peptide conformation. Model and experimental mobilities are compared by examining the data of Janini and co-workers (Janini, G.; et al. J. Chromatogr. 1999, 848, 417-433). A total of 58 peptides consisting of from 2 to 39 amino acids are considered. The average relative error between experimental and model mobilities is found to be 1.0% and the rms relative error 7.7%. In specific cases, the discrepancy can be substantial and possible reasons for this are discussed. It should be emphasized that the input parameters of the peptide model are totally independent of experimental mobilities. It is hoped that the peptide model developed here will be useful in the prediction of peptide mobility as well as in using peptide mobilities to extract information about peptide structure, conformation, and charge. Finally, we show how simultaneous measurements of translational diffusion and mobility can be used to estimate peptide charge.     

用氨基酸加和法计算多肽的脂水分配系数

Based on amino acid addition model, a set of hydrophobicity contributions of amino acids was obtained from the multivariate linear regression analysis of peptides' octanol/water partition coefficients. Multivariate regression was performed on a training set of 219 peptides including dipeptides to pentapeptides which we compose 21 natural amino acids. The correlation coefficients for the whole set fitting are 0. 978 and 0- 974, for log P and log D respectively. In addition, a new test method -evolution test-for regression analysis was discussed. The result of evolution test for amino acid addition model shows the advantage of this new test method.     

A Computer Method for Predicting the Electrophoretic Mobility of Peptides

A computer model is presented for the prediction of the electrophoretic mobilities of peptides from physical constants derived from their amino acid sequences. The model assumes that the electrophoretic mobility can be represented by a product of four functions according to the relation:  l_comp= l(L)w(W)q(Q)c(CC), where L (a length parameter) is represented by the number of amino acid residues of the peptide, W (a width parameter) is represented by the average residue mass, Q = the charge of the peptide, and CC = the position of the center of charge relative to the center of mass. The model was used to calculate the electrophoretic mobilities of peptides in a 50 mM phosphate buffer at pH 2.5. Sixty-four test peptides ranging in size from 2 to 39 amino acid residues were used for this study. The calculated mobilities show excellent correlation with experimental measurements with a correlation coefficient greater than 0.98.     

用氨基酸加和法计算多肽的脂水分配系数

在药物设计中,化合物的流水性是值得考虑的一种重要的性质．目前常使用化合物在正辛醇和水两相间的分配系数的对数值（logP）来度量其流水性．仅从化合物的结构出发来预测其脂水分配系数具有重要的意义,已有多种计算方法见诸报导’‘,“．对于普通的有机化合物,它们能给出较好的结果．多肽是一类具有重要生物功能的化合物．在药物设计中,多肽也是常用的物系．目前预测一般有机化合物脂水分配系数的方法对于多肽尚不能给出满意的结果．鉴于多肽类化合物特殊的重要性,专门发展一种方法来预测多肽的脂水分配系数也是十分必要的．在这方…     

A quantitative structure–mobility relationship of organic acids using solvation parameters

A quantitative structure–mobility relationship (QSMR) is proposed to estimate the electrophoretic mobility of diverse sets of analyses in capillary zone electrophoresis using Abraham solvation parameters of analyses, such as the excess molar refraction, polarizability, hydrogen bond acidity, basicity, and molar volume. QSMR was developed for prediction the electrophoretic mobility of 231 organic acids using the solvation parameters calculated by Abraham. Multiple linear regression (MLR) as a linear model and artificial neural network (ANN) methods were used to evaluate the nonlinear behavior of the involved parameters. The prediction results are obtained by nonlinear model, ANN, seem to be superior over MLR and were in good agreement with experimental data. In the proposed ANN–QSMR model, the overall mean percentage deviation values were 5.6, 5.4, and 5.3% and the coefficients of determinations (R²) were 0.84, 0.84, and 0.84 for training, test, and verification set, respectively. To investigate the robustness of the model, cross-validation methods have been established, i.e., leave-one-out and leave-N-out (N?=?5 and 10) and model is showed good predictive ability against data variation in cross-validation process. This model is not only able to accurately predict the migration order of a diverse set of organic acids but also model finds that solvation parameters are responsible in separation mechanism.     

Translational diffusion constants of the amino acids: measurement by NMR and their use in modeling the transport of peptides

In this work, the translational self-diffusion constants, DT's, of 12 amino acids (Ala, Arg, Asn, Asp, Cys, Glu, His, Ile, Lys, Met, Phe, and Ser) are measured by field gradient NMR and extrapolated to infinite dilution. The experiments were carried out in D2O at 298 K at pD approximately =3.5 in 50 mM sodium phosphate buffer. Of these 12 amino acids, 6 are being reported for the first time (Asp, Cys, Glu, His, Lys, and Met) and the remaining 6 (Ala, Arg, Asn, Ile, Phe, and Ser) are compared with DT's from the literature. When corrected for differences in solvent viscosity and temperature, the discrepancy between DT's measured in the present work and those reported previously is always <8%, which is reasonable given the range of values reported previously by different groups. With the present work, DT's for all of the amino acids are now available. These diffusion constants are then used in modeling studies of the diffusion and free solution electrophoretic mobility, mu, of several model peptides. For this set of peptides, it is shown that modeling using revised input parameters results in improved agreement between model and experimental mobilities.     

Using electrophoretic mobility and bead modeling to characterize the charge and secondary structure of peptides

Using a modeling methodology developed in our laboratory previously, the free solution electrophoretic mobilities of several peptides are examined to see what they can tell us about: (i) the pK(a)s of specific side groups, and (ii) possible secondary structure. Modeling is first applied to mobility versus pH data of several small peptides (Messana, I. et al., J. Chromatogr. B 1997, 699, 149) where the only adjustable parameter associated with the charge state of the peptide is the pK(a )of the C-terminal. In addition to examining this parameter, the question of possible secondary structure is addressed. For two of the peptides considered, GGNA and GGQA, it is possible to account for the observed mobilities using "random" models with little restriction on the allowed range of Phi-Psi angles. For GGRA and RPPGF, "compact" models (possibly involving an I-turn) must be used to match modeling mobilities with experiment. Finally, three more complicated peptides ranging in size from 15 to 20 amino acids are also examined and characterized (Sitaram, B. R. et al., J. Chromatogr. A 1999, 857, 263). Here also, we find evidence of I-turns or some other "compact" structure in two of the three peptides examined.     

Determination of dissociation constants of aristolochic acid I and II by capillary electrophoresis with carboxymethyl chitosan-coated capillary

Aristolochic acid-I and aristolochic acid-II have been proved to be the main bioactive and toxic component in Aristolochia plants. As a result, the determination of their dissociation constants, which are important property parameters for weak acids, is highly desired for related pharmacological and toxicological studies. In this work, the dissociation constant values of aristolochic acid-I and aristolochic acid-II were determined by capillary electrophoresis using carboxymethyl chitosan-coated capillary, based on their electrophoretic mobilities by using nonlinear regression as well as linear regression, showing that the two models give comparable results. The data were also compared with those obtained by capillary electrophoresis with polybrene-coated capillary, and no conspicuous difference was observed. The correlation coefficients were all higher than 0.998 for both linear and nonlinear regression model. The pKa values were found to be 3.3±0.1 for aristolochic acid-I and 3.2±0.1 for aristolochic acid-II.     

Determination of dissociation constants of amino acids by capillary zone electrophoresis.

A method is proposed for the determination of dissociation constants of amino acids by capillary zone electrophoresis. According to the dissociation equilibrium of amino acids and the basic theory of electrophoresis, the nonlinear relationship between the pH value of the buffer and the effective electrophoretic mobilities of the analyte was obtained. The dissociation constants can be calculated from the pH values and the corresponding effective electrophoretic mobilities using the program written in C++. The dissociation constants, pKa1 and pKa2, of 11 kinds of amino acids were determined successfully by the proposed method. The determined dissociation constants were compared with values in the literature; the differences between them are in the range of -0.03 to 0.06. No significant differences were observed between the determined dissociation constants and the corresponding literature values.