Modelling the solubility of the 20 proteinogenic amino acids with experimentally derived saturation data

Nathan A. Bowden

Research output: Thesisinternal PhD, WU

Abstract

In this thesis, experimental work and thermodynamic modelling is presented. The application of this is the extraction of proteins from agro-industrial residue streams, the hydrolysis of the proteins to their constituent amino acids and the separation and crystallization of amino acids. The development of this process will benefit the reduction of waste and the worlds reliance on non-renewable resources.

Crystallizing individual α-amino acids from a mixture of α-amino acids would improve their separation in a bio-refinery, but the solubility of the 20 α-amino acids in a mixture of 20 α-amino acids is unknown. In order to understand the solubility of α-amino acids in mixtures of 20 α-amino acids, we first studied the solubility of the 20 α-amino acids in single solution using the van ‘t Hoff and Sober equations and data available in the literature. In chapter 2, the results of these equations were compared and their coefficients are reported. Then we measured the solubility of the 20 α-amino acids in a model system of an industrial residue containing all 20 α-amino acids and found that only l-tyrosine, l-tryptophan, l-aspartic acid, l-glutamic acid and l-asparagine dissolved in higher concentrations in the model system of 20 α-amino acids than in aqueous solution by themselves. The solubility of all of the aliphatic α-amino acids was lower in the mixture. The longer the side chain length of the aliphatic α-amino acids, the greater the decrease in solubility in the mixture as compared to being in solution by itself. The maximum solubility of l-lysine and l-glutamine showed the most decrease when in the mixture in comparison to in aqueous solution alone. Our results demonstrate that α-amino acids with similar physical structures have similar changes in solubility when in a mixture of α-amino acids. We assert that α-amino acids can be grouped by their physical structure.

In chapter 3, we show that the addition of organic solvents to α-amino acids in aqueous solution could be an effective method in crystallization. The available data on the solubility of α-amino acids in water, water-ethanol mixtures and ethanol at 298.15 K and 0.1MPa were reviewed. The solubility of L-alanine, L-proline, L-arginine, L-cysteine and L-lysine in water and ethanol mixtures and the solubility of L-alanine, L-proline, L-arginine, L-cysteine, L-lysine, L-asparagine, L-glutamine, L-histidine and L-leucine in pure ethanol systems were measured and are published here for the first time. The impact on the solubility of amino acids that can convert in solution, L-glutamic acid and L-cysteine, was studied. At lower concentrations, only the ninhydrin method and the UPLC method yield reliable results. In the case of α-amino acids that convert in solution, only the UPLC method was able to discern between the different α-amino acids and yields reliable results. The results demonstrate that α-amino acids with similar physical structures have similar changes in solubility in mixed water/ethanol mixtures. The solubility of L-tryptophan increased at moderate ethanol concentrations.

Since the knowledge on the effects of organic solvents on the solubility of α-amino acids is incomplete, the intermolecular forces between the α-amino acids are not fully understood. In chapter 4, the solubilities of each of the 20 proteinogenic α-amino acids in a mixture of all the 20 proteinogenic α-amino acids in 0%, 15%, 30%, 50%, 70% and 80% (g/g) ethanol-water solutions and at 277.15, 297.15, 315.15 and 335.15 K are reported and compared to literature values of single α-amino acids. The solubility of the individual α-amino acids in water, ethanol and water-ethanol mixtures were discussed in chapter 3 and are compared with this data. The results show that amino acids can be grouped according to the structure of their side chains. If branched aliphatic, hydroxylic, phenylic and carboxylic groups are on the side chain of an amino acid, then the solubility of that amino acid will increase when in a mixture of 20 amino acids between 30-70% (g/g) ethanol-water solution. If sulphur containing and amine/amide containing groups are on the side chain of an amino acid, then the solubility of that amino acid will decrease in a mixture of 20 amino acids between 30-70% (g/g) ethanol-water solution compared to the solubility as a single amino acid.

In chapter 5, the solubility data of all proteinogenic α-amino acids in binary ethanol/water systems is used to model their excess solubility. The empirical and regressive models of Gude and NRTL and the predictive Jouyban-Acree model are applied. Based on the results, it is hypothesize that amino acids that are spherical and lack a reactive side chain show little or no excess solubility. Being rod-like and/or having a reactive side chain leads to a positive excess solubility in a mixed solvent of ethanol and water. The empirical and regressed models, NRTL and Gude, fit the data well and the predictive Jouyban-Acree model, not originally intended to be used for small molecules, is less accurate but offers insights into the thermodynamic properties of the amino acids.

In chapter 6, the solubility data of all proteinogenic α-amino acids in binary ethanol/water systems to model their excess solubility is used again but use it to validate a novel predictive model. In order to predict α-amino-acid solubility in a solution of water and ethanol, the van Laar equation for the molar excess Gibbs energy is used. To obtain meaningful activity coefficients from the solubility data using the van Laar equation, the ratio of the fugacity of the solid α-amino acid to that of the subcooled liquid amino acid is needed. That ratio is obtained from estimated melting temperatures and enthalpies of fusion. The ternary van Laar equation provides a predictive model for obtaining the solubility of an α-amino acid in an ethanol-water solvent. The normalized root mean square variances (NRMSV) for 16 of the 20 solubility predictions are below 0.100, indicating very good agreement with the solubility data. The NRMSV of the other four predictions are below 0.220, indicating good agreement with the α-amino-acid solubility data. Six of the 20 amino acids could be calculated using previously published group contribution data. Additional group contribution data is reported here for seven amino acids.

Amino acids can be obtained from protein after hydrolysis. In addition, several agro-industrial residues already contain a mixture of free amino acids. The objective of chapter 7 was to develop a method for amino acids separation, starting from mixtures containing amino acids, and using anti-solvent precipitation with ethanol. Protamylasse™, rubber seed protein hydrolysates and grass juice were used in the experiments, representing existing and potential agro-industrial residues. The results show that in a water-ethanol system, some amino acids had lower solubility in mixtures than as a single component, thereby facilitating precipitation. A sufficiently high total amino acid concentration in the mixture is needed to achieve precipitation, therefore a concentration step is sometimes required. Ethanol precipitation can be applied as a pre-treatment to separate mixtures into groups of amino acids or a polishing step to increase purity.

In conclusion, this thesis provided novel insights in protein extraction, protein hydrolysis, amino acid separation and the use of models for solid-liquid interactions. In this way, future researchers benefit not only from the data and protocols developed from this research but also the knowledge gained from its application.

Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Wageningen University
Supervisors/Advisors
  • Sanders, J.P.M., Promotor
  • Bruins, Marieke, Co-promotor
Award date16 May 2018
Place of PublicationWageningen
Publisher
Print ISBNs9789463438612
DOIs
Publication statusPublished - 2018

Fingerprint Dive into the research topics of 'Modelling the solubility of the 20 proteinogenic amino acids with experimentally derived saturation data'. Together they form a unique fingerprint.

Cite this