»¯Ñ§ºÏ³ÉÉúÎӢÎİ棩£º05618_08

231 Industrial production of amino acids by fermentation and chemo-enzymatic methods 8.1 Introduction 8.2 Essential and nonessential amino acids 8.3 Stereochemistry of amino acids 8.4 Amino acid fermentation 8.5 Recovery of the amino acid from the fermentation broth 8.6 Case study: fermentative production of L-phenylalanine from glucose 8.7 Case study: The production of L-phenylalanine by enzymatic methods Summary and objectives Appendix 8.1: Representation of and the nomenclature for stereeisomers Appendix 8.2 Examples of industrial production of amino acids by A8.2.1 Enzymatic resolution of racemates enzymatic methods A8.2.2 Enzymatic asymmetric synthesis A8.2.3 Oxidation - Reduction reactions 232 234 236 240 248 253 262 272 273 277 277 286 289 232 Chapter 8 essential ammo adds non-pcoteino- genic amino acids precursors for me chemical synthesis Industrial production of amino acids by fermentation and cherno-enzymatic methods 8.1 Introduction Amino acids have always played an important role in the biology of life, in biochemistry and in (industrial) chemistry. There are several reasons why they are of commercial interest. Firstly, amino acids are the building blocks of proteins and they play an essential role in the reguiation of the metabolism of living organisms. Largescale chemical and microbial production processes have been commercialised for a number of essential amino acids. The use of glutamic acid, lysine and methionine as food and feed additives is well established nowadays. Secondly, current interest in developing peptide-derived chemotherapeutics has heightened the importance of rare and non-proteinogenic pure amino acids. For example, D-phenylglycine and D-phydroxy-phenylglycine are building blocks for the broad spectrum f3-ladam antibiotics ampicillin and amoxycillin, respectively. The natural amino acid L-valine is used as feedstock in the fermentative production of the cyclic peptide cyclosporin A, which has immuno-suppressive activity and is used in human transplant surgery. Thirdly, amino acids are versatile chiral (optically active) building blocks for a whole range of fine chemicals. In the last two decades, there has been a growing public awareness and concern with regard to the exposure of man and his environment to an ever increasing number of chemicals. The benefits, however, arising from the use of therapeutic agents, pesticides, food and feed additives, etc are enormous. Hence there is still an ever increasing demand for more selective drugs and pesticides which are targeted in their mode of action, exhibit less toxic side-effects and are more environmentally acceptable. To this end a central role will be played by chiral compounds, as nature at the molecular level is intrinsically chiral. Consequently, this provides an important stimulus for companies to market chiral products as pure optical isomers. This in turn results in an increasing need for efficient methods for the industrial synthesis of optically active compounds. Amino acids are, therefore, important as nutrients (food and feed), as seasoning, flavourings and starting material for pharmaceuticals, cosmetics and other chemicals. They can be produced in a variety of ways (see Table 8.1): 0 chemical synthesis; 0 isolation from natural materials (extraction); amino acid fermentations (using micro-organisms); 0 chemo-enzymatic methods. In this chapter we consider amino acid production by fermentation and by chemoenzymatic methods. We first consider the stereochemistry of amino acids and the importance of chirality in chemical synthesis. General approaches to amino acid fermentation and recovery of amino acids from fermentation broths are then dealt with, followed by a detailed consideration of the production of L-phenylalanine by direct fermentation. Later in this chapter, chemo-enzymatic methods of amino acid Industrial production of amino acids by fermentation and chemo-enzymatic methods 233 fennentation are examined. We first consider general aspects of the approach, followed by more detailed case studies. We have again selected L-phenylalanine for detailed consideration, since this important amino acid can be produced by direct fermentation and by a variety of chemo-enzymatic methods. This allows comparisons between the two different approaches to be made, including a consideration of economic aspects of large scale production of the amino acid. Two appendices are included at the end of this chapter. The first is intended to serve as a reminder, for those of you who might need it, of the nomenclature and representation of stereoisomers. The second appendix contains descriptions of various chemo-enzymatic methods of amino acid production. This appendix has been constructed largely from the recent primary literature and includes many new advances in the field. It is not necessary for you to consult the appendix to satisfy the learning objectives of the chapter, rather the information is provided to illustrate the extensive range of methodology associated with chemo-enzymatic approaches to amino acid production. It is therefore available for those of you who may wish to extend your knowledge in this area. Where available, data derived from the literature are used to illustrate methods and to discuss economic aspects of large-scale production. + amino ackl chemical extradlon fermentation enzymatic synthesls CatalySlS L-alanine + + L-arginine + + L-aspartic acid + + L-cystine + L-glutamic acid (Na) (+I + L-histidine (.HCI) + + L-isoleucine + + L-leucine + L-lysine (.HCI) + + L-methionine + L-phenylalanine (+) (+) + L-proline + (+I L-serine + + L-threonine + + L-tryptophan + + L-tyrosine + L-valine + (+) + L-cysteine Table 8.1 Production methods of proteinogenic amino acids. 234 Chapter 8 8.2 Essential and nonessential amino acids a-carbon Lconfiguration dimer An amino acid is defined as a compound that possesses both amino and carboxyl groups. Some amino acids are iminocarboxylic acids, such as proline while others are sulphur containing amino acids, such as cysteine and methionine (Table 8.2). Over 100 amino acids have been isolated and identified from natural sources to date. The great majority of these naturally occumng amino acids have the amino group attached to the carbon a to the carboxylic acid. With very few exceptions, the a-carbon also bears a hydrogen atom. The fourth bond of the a-carbon is pined to a group which has over 100 variations. Thus, most of the naturally occurring amino acids differ only in the structure of the organic residue attached to the a-carbon. An interesting and important fact is that almost all amino acids isolated from proteins have the L-configuration at the a-carbon, although some amino acids isolated from microbiological sources are the mirror image isomers, ie in the Dconfiguration. We shall consider amino acid steremhemistry in more detail in section 8.3. Of the amino acids isolated from living material, only about 20 are naturally Occurring components of proteins. Some of these are shown in Table 8.2. The remainder, non-proteinogenic amino acids, are found as intermediates or end products of metabolism. One of the amino acids commonly found in protein hydrolysates is called cystine; it has the following structure: HOOC- CH-CH,-S-S-CH,-CH-COOH (CYSCYS) I I NH2 NH2 cystine It is clearly a dimer of cysteine, where the thiol groups have been oxidised to form a disulphide linkage. The dimer actually results because of two monomers at widely spaced intervals in the polypeptide are joined together by a disulphide bridge. Thus the basic amino acid is cysteine and consequently, the dimer is not included here. Industrial production of amino acids by fermentation and chemo-enzymatic methods 235 Table 8.2 structure of some proteinogenic a-amino acids. All living species are able to synthesise amino acids. Many species, however, are deficient in their ability to synthesise within their own metabolic system all the amino acids necessary for life. The eight amino acids with this special significance for the human species are called essential amino acids, these are: L-valine; 0 L-leucine; L-isoleucine; 0 L-threonine; 0 L-methionine; L-phenylalanine; 0 L-tryptophan; L-lysine. They are essential not because they are the only amino acids requred for human functioning, but because they are essential in the diet of the human species. essential amino acids 236 Chapter 8 1) Name thee sulphur containing amino acids. 2) Name five of the eight essential amino acids. 3) Name two amino acids that contain a heterocyclic ring. 4) Name the amino acid with the simplest structure. 5) Name the amino acid considered to be a dimer. 6) Name an amino acid that is produced industrially only by enzymatic 7) Name an amino acid that is produced industrially only be chemical catalysis. synthesis. 8.3 Stereochemistry of amino acids The French physicist Biot discovered during the early nineteenth century, that a number of naturally occumng organic compounds rotate the plane of polarisation of an incident beam of polarised light. In the latter part of the nineteenth century, it was found that many pairs of compounds seemed to have an identical structure and identical physical properties, such as melting point and solubility. Compounds in each pair were differentiated by the fact that even in solution they rotated polarised light in equal amounts but in opposite direction. Such compounds are called optical isomers and are described as being optically active. Optical activity requires, and is explained by an asymmetric arrangement of groups around a tetrahedral carbon atom. The geometric properties of a tetrahedron are such that if there are four different substituents attached to a carbon atom, the molecule does not contain a plane of symmetry, and there are two kinds of geometrical arrangements which the molecule can have. These two arrangements (configurations) are different in that it is not possible to simultaneously superimpose all the atoms of one figure on the like atoms of the other. The two configurations are, in fact non-superimposable mirror images. An illustrative example is given in Figure 8.1. Optical isomers mra~~edra~ cabon a~m Industrial production of amino acids by fermentation and chemo-enzymatic methods 237 I I Figure 8.1 Non-superimposable mirror images. Such molecules result when the four groups attached to the carbon atom are all different and a molecule of this kind is said to be asymmetric, or to contain an asymmetric carbon. Molecules that are not superimposable on their mirror images are chiral. If two compounds are related as non-superimposable mirror images, they are called Can you explain why the amino acid alanine is optically active, whereas glycine n is not (refer to Table 8.2)? We can see from Table 8.2 that the a-carbon of alanine is asymmetric (four different groups attached), whereas that of glycine is not. Optical activity requires an asymmetric carbon atom. If two enantiomers are mixed together in equal amounts the result is a racemic mixture. We meet a number of enantiomeric items in daily life. The left hand, for example, is the mirror image of the right hand and they are not superimposable (see Figure 8.1). This becomes obvious if we try to put a right glove on a left hand. Similarly, a pair of shoes is an enantiomeric relationship while the stock in a shoe store constitutes a racemic mixture. Representation of and the nomenclature for stereo-isomers are given in Appendix 1. 8.3.1 Importance of chirality If we consider natural synthetic processes, enzymes are seen to exert complete control over the enantiomeric purity of biomolecules (see Figure 8.2). They are able to achieve this because they are made of single enantiomers of amino acids. The resulting enantiomer of the enzymes functions as a template for the synthesis of only one enantiomer of the product. Moreover, the interaction of an enzyme with the two enantiomers of a given substrate molecule will be different. Biologically important molecules often show effective activity as one enantiomer, the other is at best ineffective or at worst detrimental. asymmetric carbon enantiomem enantiomers. racemic mixtures enantiomeric PJw 230 Chapter 8 Figure 8.2 Enzyme interaction with two enantiomers of a given substrate molecule. In some cases the unwanted enantiomer can perturb other biological processes and cause catastrophic side effects. The use of enantiomerically pure compounds thus permits more specific drug action and the reduction in the amount of drug administered. Even in the cases where the other enantiomer is inactive, the work involved in its metabolism before secretion can be avoided. Numerous examples of the different biological effects of enantiomers are available. One of the enantiomers of limonene smells of lemons, the other of oranges; one of carvone smells of caraway, the other of spearmint. These differences obviously have important sped:iz Industrial production of amino acids by fermentation and chemo-enzymatic methods 239 consequences for the perfume and flavour industries. Both enantiomers of sucrose are equally sweet, but only the naturally occurring D-enantiomer is metabolid, making the synthetic L-enantiomer a potential dietary sweetener. In the protection of mps from insects, one enantiomer of a compound may be a repellant while the other is an attractant, and the racemic mixture is ineffective. One enantiomer of penicillamine (D-) exhibits antiarthritic properties but the other is highly toxic (Figure 8.3). The teratogenic effects of thalidomide were induced by one enantiomer, the other exhibited the beneficial effects against morning sickness. Different optical enantiomers of amino acids also have different properties. L-asparagine, for example, tastes bitter while D-asparagine tastes sweet (see Figure 8.3). L-Phenylalanine is a constituent of the artificial sweetener aspartame (Figure 8.3). When one uses D-phenylalanine the same compound tastes bitter. These examples clearly demonstrate the importance of the use of homochiral compounds. hvnce Of hanochlral compo,,,,^ ~ ~~~ ~~ Figure 8.3 Examples of different biological effects of enantiomers. S and R refer to a particular system of nomenclature used to describe chiral carbon. (see Appendix A8.1) List possible advantages of using enantiomerically pure compounds as drugs, as opposed to using racemic mixtures. 240 Chapter 8 8.4 Amino acid fermentation wild strains Many micro-organisrns accumulate amino acids in culture media. Indeed, wild strains have proved to be effective producers of amino acids like alanine, glutamic acid and valine. Since amino acids are used as essential components of the microbial cells and their biosynthesis is regulated to maintain an optimal level, they are normally synthesised in limited amounts and are subject to negative feedback control. The main problem using wild strains is, therefore, the production of minor amounts of amino acids at an early stage in the fermentation, giving rise to feedback control. To achieve overproduction of amino acids the following procedures can be used: 0 improvement of the uptake of the raw material (starting material); 0 hindrance of the side reactions; tzgz mw 0 stimulation of the enzymes that are involved in the synthesis; 0 inhibition of the degradation of the desired amino acid; 0 stimulation of excretion of the amino acid that is produced. The most successful way to achieve overproduction is to make use of mutants. Another way to overcome feedback rrgulation is to make use of a kind of semi-fermentation process called precursor addition fermentation; this will be considered later in this chapter. Amino acids produced by fermentation on an industrial scale are listed in Table 8.3. smegies br overpcoddm amino aclds tonneelyear aspartic acid ca. 8,000 glutamic acid lysine phenylalanine threonine tryptophan ca. 270,000 ca. 90,000 ca. 8,000 ca. 500 ca. 100 applications aspartame (sweetener) enzymatic synthesis of alanine and phenylanine flavours, pharmaceuticals dietary aspartame (sweetener) dietary pharmaceuticals, dietary Table 8.3 Amino adds industrially produced by fermentation. Examine the list of p'ocedures to achieve overproduction (shown above) and n idenbfy which ones could be achieved by mutation of a wild strain. Since all the procedures listed involve enzymes they all could be achieved by mutation. This emphasises the potential of using mutation for amino acid production. Industrial production of amino acids by fermentation and chemo-enzymatic methods 241 8.4.1 Wild strains Normally amino acid synthesis will just satisfy the metabolic demand. In some cases, when the amino acid occurs in both biosynthetic and energy production pathways, overproduction of the amino acid can take place. This is the case, especially for L-glutamic acid, with Corynebacterium, Brm'bacterium, Microbacterium and Arthbacter. These micro-organisms have some common features: Gram-positive, non-spore forming, non motile, cocci or rod and biotin-requiring for growth. The following general characteristics have been found to influence the overproduction of L-glutamic acid: characteristic influmdng ove~duclian 0 nutritional requirement for biotin; 0 intracellular levels of phospholipids; 0 lack of, or low activity of, a-oxoglutarate dehydrogenase. In the first two cases, the permeability of the cell membrane to L-glutamate is altered through changes in the fatty acid composition of the cell membrane. In the third case, the degradation of the amino acid is inhibited, resulting in accumulation. Other amino acids produced by wild strains include L-valine, DL-alanine and L-proline. n Why are amino acid synthesised in only small amounts in wild strains? The small overproduction of amino acids by wild type strains in culture media is the result of regulatory mechanisms in the biosynthetic pathway. These regulatory mechanisms are feedback inhibition and repression. These mechanisms are considered fully in other BIOTOL texts. However, details of the mechanisms are not essential for an understanding of the material in this chapter. In this section we will, therefore, only briefly desaibe the mechanisms. Feedback inhibition In the metabolic pathway to an amino acid several steps are involved. Each step is the result of an enzymatic activity. The key enzymatic activity (usually the first enzyme in the synthesis) is regulated by one of its products (usually the end product, eg the amino acid). If the concentration of the amino acid is too high the enzymatic activity is decreased by interaction of the inhibitor with the eatery site of the enzyme (allosteric enzyme). This phenomenon is called feedback inhibition. Repression In repression the enzyme concentration is regulated at the DNA level. The genetic information for a certain enzyme is located on the genetic material in the cell, the DNA. In general this consists of several integrated regions working together (an operon). 242 Chapter 8 If the information can be translated into messenger RNA, and this information can be transferred to the ribosome, the enzyme can be synthesised. If not, the key enzyme in a metabolic route to the wanted amino acid is not synthesised. This might involve a repressor, consisting of an apo-repressor (ie a protein from a regulatory gene) and the wrepressor (normally the end product of the pathway) which binds to the operator gene and prevents translation of the operon. Normally this is not an absolute effect. If the concentration of the end product is high the end product blocks the operator gene, so the m-RNA cannot be formed and the enzyme cannot be produced. The concentration of the enzyme is kept low in such a case. It is obvious that in the case of overproduction of amino acids the above mentioned regulatory mechanisms are not wanted. One way to overcome these regulations is to make use of mutants. 8.4.2 Mutant strains --repressor and mWr-r Two types of mutants have been used for amino acid overproduction: auxotrophic and regulatory mutants. In some cases, mutant strains have been further improved through DNA-recombination. Auxotrophic mutants Auxotrophic mutants are mutants that miss one or more of the enzymes used in the biosynthetical pathway for one or more amino acids. In practice this means that the mutant needs one or more key metabolites which it cannot synthesise for growth in its growth medium. For example, consider Figure 8.4. Figure 8.4 Skeleton pathway leading to Lphenylalanine, tyrosine and tryptophan in Escherichia coli. In the case of a tyrosine auxotrophic mutant, the mutant does not produce at least one of the enzymes to synthesise tyrosine (E6 in Figure 8.4). In the literature this is described as Atyr. This means that a small amount of tyrosine has to be added to the culture medium, just enough to support growth, because the micro-organism is not capable of producing tyrosine. In the wild strain, El is controlled by feedback inhibition involving tyrosine. So, accumulation of tyrosine slows the rate of flow through the pathway. In the mutant strain, however, this does not occur because tyrosine is not produced, so overproduction of phenylalanine will occur. The genetic marker of this particular organism is reported in literature as Tyr-. phmylalanine ove~*clim in Tyf Industrial production of amino acids by fermentation and chemo-enzymatic methods 243 branched pa*wars selection using toxic anabgues regulation by and tryptophan tyrosine. phenylatanine Auxotrophic mutants are used in the production of end products of branched pathways, ie pathways leading to more than one amino acid at the same time. This is the case for L-lysine, L-methionine, L-threonine and Lisoleucine in Brertibacterium flmm and Corynebacterium glutamicum. Regulatory mutants There are two possible sites that are genetically inactivated in regulatory mutants: the regulatory site; the functions of repressor and wrepressor. These mutants lack feedback inhibition and are used for the production of many amino acids. Selection of these regulatory mutants is often done by using toxic analogues of amino acids; for example p-fluoro-DL-phenylalanine is an analogue of phenylalanine. Mutants that have no feedback inhibition or repression to the amino acid are also resistant to the analogue amino acid. They are therefore selected for and can be used to overproduce the amino acid. Some amino acid analogues function as false co-repressors, false feedback inhibitors or inhibit the incorporation of the amino acid into the protein. The best amino acid producers are organisms that are both auxotrophic and regulatory mutants. What advantage do regulatory mutants have, when compared to auxotrophic n mutants, for amino acid fermentation? Unlike auxotrophic mutants, regulatory mutants can be grown in inexpensive, complex media and they do not require careful control of growth conditions. DNA-recombinant micro-organisms and combinations Another way to enhance the production of an amino acid is to make use of DNA-recombinant technology, often in combination with the mutations already described. In this way the negative features of the miaworganisms are avoided. To help explain this, we will consider a well known fermentation of L-phenylalanine using Escherichia coli. We have already seen the metabolic pathway leading to the production of L-phenylalanine in Figure 8.4. Conversion of erythrose-4-P and phosphoenol pyruvate to 3-deoxy-D-arabinoseheptulosonic acid (DAW) is catalysed by DAHP synthetase. In E.coli there are three isoenzymes of the enzyme; these are known as mF (rrgulated by tyrosine), moG (regulated by phenylalanine) and mH (regulated by tryptophan). In each case, regulation is both at the level of enzyme formation (repression) and enzyme activity (feedback inhibition). Another site regulated through phenylalanine controls the expression of the structural genes (pheA) for chorismate mutaseprephrenate dehydratase (E3 and E5 in Figure 8.4). The mechanism is based on the phenylalanine mediated repressor from the pheA regulator gene that binds to the pheA operator site. The effect of this regulation (on the pheA level) and feedback inhibition (on the moG level) is low levels of the enzyme chorismate mutase-prephrenate dehydratase and low activities of the enzyme DAHP synthetase. This results in low levels of phenylalanine. 244 Chapter 8 To achieve overproduction of phenylalanine, the micro-organism should be drmpressed at the pheA level and free of inhibition at the ad level. Both genes are located on the chromosomal DNA of the micro-organism and, by means of amino acid analogues such as pfluomDLphenylalanine it is possible to make (phenylalanine) feedback resistant mutants of E.di (pheA and mp mutants). The following procedure can be used: the genes are isolated from the chromosomal DNA and put into a plasmid (circular regulation of the pkAR structural gene in the plasmid is altered by using a strong promoter (several very strong promoters are nowadays used, eg the lac promoter); 0 the total effect (ie overproduction of phenylalanine by deregulation) can be enhanced by using more than one pheAFR genes in the plasmid, in a so called multicopy recombinant. leedbadc resistant mutants d use of plasma Dm extrachromosomal piece of DNA); In a good production strain, the plasmid should be pmnt in a stable way and should not be lost from the micro-organism after a few generations. The pdure described above is just one way to get to a good phenylalanine n production strain. Briefly outline another way to achieve the goal. The most obvious alternative approach is to deregulate the mF gene, which is subject to tyrosine regulation. This, of course, could be achieved by DNA-recombinant techniques or by mutation. Provide brief explanations for the following 1) Wild strains of E. coli are not used for L-phenylalanine production by direct 2) Auxotrophic mutants of E. coli are particularly useful for the production of fermentation. L-phenylalanine by direct fermentation. E. wZi. 3) Regulatory mutants improve the rate of L-phenylalanine overproduction by 4) An E. coli mutant, Tv-, is likely to enhance L-phenylalanine overproduction. 8.4.3 Methods of fermentation The growth of micro-organisms used in the production of amino acids is done in a well balanced environment. The conditions required are: a controlled pH of the fermentation medium (approximately neutral); richgrowthmedia; highly aerobic conditions; sterile conditions. fermenta~on conditions Industrial produdion of amino acids by fermentation and chemo-enzymatic methods 245 methods of amino add fermentation short fermentation time preferable bw residual substrate concentration control of oxygen demand advantages of continuous fermentation Slight changes in the fermentation conditions can greatly affect amino acid produdion. These variations are sometimes caused by changes in sterilising conditions, agitation and aeration, temperature, pH, pressure and liquid level. This is why the parameters have to be controlled in a proper way using precise equipment. We can differentiate between three possible methods of amino acid fermentation: batch fermentation; fed-batch fermentation; continuous fermentation. 1) Batch fermentation Batch fermentation is the most widely used method of amino acid production. Here the fermentation is a closed culture system which contains an initial, limited amount of nutrient. After the seed inoculum has been introduced the cells start to grow at the expense of the nutrients that are available. A short adaptation time is usually necessary (lag phase) before cells enter the logarithmic growth phase (exponential phase). Nutrients soon become limited and they enter the stationary phase in which growth has (almost) ceased. In amino acid fermentations, production of the amino acid normally starts in the early logarithmic phase and continues through the stationary phase. For economical reasons the fermentation time should be as short as possible with a high yield of the amino acid at the end. A second reason not to continue the fermentation in the late stationary phase is the appearance of contaminant-products, which are often difficult to get rid off during the recovery stage. In general, a relatively short lag phase helps to achieve this. The lag phase can be shortened by using a higher concentration of seed inoculum. The seed is produced by growing the production strain in flasks and smaller fermenters. The volume of the seed inoculum is limited, as a rule of turd normally 10% of the fermentation volume, to prevent dilution problems. 2) Fed-batch fermentation Fed-batch fermentations are batch fermentations which are fed continuously, or intermitantly, with medium without the removal of fluid. In this way the volume of the culture increases with time. One of the advantages of the fed-batch fermentation is the fact that the residual substrate concentration may be maintained at a very low level. This may result in a removal of catabolite repressive effects and avoidance of toxic effects of medium components. Another advantageous effect is on the oxygen balance. The feed rate of the carbon source (mostly glucose) can be used to regulate cell growth rate and oxygen limitation, especially when oxygen demand is high in the exponential growth phase. 3) Continuous fermentation In general, there are several advantages of continuous fermentation as compared to batch fermentations. These include higher productivity, operation for a very long period of time, and lower installation and maintenance costs. The maintenance costs are particularly important. In batch cultures, oxygen demand, pH control requirements and amount of cooling required, changes throughout the whole fermentation run, whilst in continuous fermentations these factors are constant. 246 Chapter 8 produclion strain variability multi-stage axltinWUS sysOsm A disadvantage of using continuous fermentation is the chance of contamination by other micmrganisms during the long fermentation runs (sometimes several weeks). Although this sometimes happens, it should not be overemphasised, since most of the time the fermentation conditions are such that a special niche is created, and only a limited number of other micro-organisms can grow. A more serious problem is the occurrence of variants of the parent production strain by back mutation or loss of genetic elements (plasmids). In addition, phage infections are more likely to occur. Despite the advantages of continuous cultures, the technique has found little application in the fermentation industry. A multistage system is the most common continuous fermentation and has been used in the fermentation of glutamic acid. The start-up of a multi-stage continuous system proceeds as follows. Initially, batch fermentation is commenced in each vessel. Fresh medium is introduced in the first vessel, and the outflow from this proceeds into the next vessel. The overall flow rate is then adjusted so that the substrate is completely consumed in the last vessel, and the intended product accumulated. The concentration of cells, products and substrate will then reach a steady state. The optimum number of vessels and rate of medium input can be calculated from simple batch experiments. 8.4.4 Problems in the fermentation of amino acids The following problems in the industrial fermentation of amino acids may arise: 0 contamination of the culture by other microaganisms during fermentation; 0 bad fermentation reproducibility due to differences in raw material; back mutation or loss of genetic material of the production strain; 0 infection of the culture by bacterial viruses (phages). Although the first two possibilities can lead to severe problems in the fermentation of amino acids, these problems can be prevented by: using proper plant design; maintenance of hygienic conditions throughout the operation; reservation of large batches of raw material with uniform qualities. Much more severe (and much more difficult to control) are the last two possibilities which will now be discussed in more detail. Back mutation and loss of genetic material Back mutation (reversion) of the mutant production strain to the wild type (prototrophic form) can occur. Since auxotrophic mutants and regulatory mutants are widely used in the overproduction of amino acids, this can be a severe problem. In ~tuI.e, mutation always takes place but this takes some time. However, in fermentation many generations are produced in a relatively short period of time and the chances of back mutation are enhanced. n Make a list of possible ways of solving the problem of reversion. The main ways to solving the problem of reversion are: Industrial production of amino acids by fermentation and chemo-enzymatic methods 247 soluthm he 0 make use of fresh starting material (inoculum) for each run; problem of 'reversian' 0 make use of antibiotics that inhibit growth of the wild type (or revertant) but not the mutant; 0 start with mutants which are very stable (many companies spend a lot of money in the isolation of stable mutant strains). Loss of genetic material (for example the constructed plasmid containing the genetic information necessary for overproduction of the amino acid) can also occur. This will happen in situations where there is an energetically favourable advantage for the mirro-organism not to have the plasmid in the cell. Possible ways to overcome the problem are: construction of a strain in such a way that it is energetically advantageous to overproduce the requid amino acid, thus keeping the construct in the cell; introduction of a genetic marker into the construct (usually antibiotic resistance encoded by plasmid genes) and operation of the fermentation under selective pressure by adding the relevant antibiotic to the medium. Only the micro-organisms carrying the marker would survive, while others die off. Phage infection Phage infections sometimes occur in big fermentation plants and cause severe damage. Phages are bacterial viruses composed of a nucleic acid core (DNA or RNA) in a protein envelope. Infection proceeds in the following way: adsorption onto the bacterial cell followed by introduction of genetic material into the bacterium; stages in Phage inkdm multiplication within the cell making use of the genetic apparatus of the bacterial cell; liberation of the infective phages by lysis of the cell resulting in the death of the bacterial cell. There are several different groups of phages, each with there own charackristics and host range specifiaty. Phages are widely distributed in nature (soil, air, raw materials etc), they are rather stable and can be introduced to the fermentation easily through the air. Ph infections can be recognised if fermentation characteristics change. The slowing T own of the p'ocess (for example delay in onset of the fermentation) and decmase in amino add production yield are the first signs of a moderate form of phage infection. If lysis of the cells occu~s in the growth phase, accompanied by a rise in pH and increase in fuaming, this suggests that there is a severe phage infection. confirlnation of a phage infection can be done by electron microscopic observation. 248 Chapter 8 An additional problem is the presence of latent forms of phage infections, in which the production strain carries the phage genome integrated in the chromosome and only in specific situations (after induction) releases the phages. These may then be changed to the virulent form and infect other micro-organisrns present. Prevention of phage infections is difficult and there is no absolute preventative measure. However, phages have their own optimum conditions (pH and temperature) for infection and replication. By choosing the proper conditions it is possible to create an environment that is favourable for bacterial growth but not for phage multiplication. Other ways to prevent phage contamination or multiplication are: 0 plant hygiene (sterilisation of raw material; disinfection of equipment by temperature or chemical agents; using proper air filters, etc); 0 isolation of phage resistant strains - however, the problem is that after a while other types of phages become capable of infecting the strain; 0 inhibition of phage multiplication by prevention of phage adsorption to the cell wall using chemical agents during fermentation (chelating agents capture calcium ions necessary for adsorption; use of surfactants like Tween). latent phage infections Give possible reasons for selecting each of the following operating conditions for a hypothetical amino acid fermentation. 1) Fed-batch mode. 2) Monoseptic. 3) Antibiotics added to medium. 4) Fresh inoculum used for each run. 5) Chelating agent added to medium. 8.5 Recovery of the amino acid from the fermentation broth Methods of product recovery are considered in detail in the BICKOL text entitled 'Product Recovery in Bioprocess Technology'. In this chapter we will briefly review the methods applicable to recovery of amino acids. General criteria for the economically feasible recovery of amino acids from the generalaisria SaSibiMy f~r~~ic fermentation broth are: the recovery should be as simple as possible; the recovery yield of the processes should be high; process steps that give rise to loss of product should be avoided as much as possible; the process shouid be easy to scaleup industrially. Industrial production of amino acids by fermentation and chemo-enzymatic methods 249 Amino acid fermentations do have several advantages over other kinds of fermentations, like for example the production of enzymes, in respect of recovery of desired product. Make a list of the advantages that you can think of. n The main advantages are: 0 normally the production strain is constructed in such a way that overproduction of the desired amino acid is obtained and no, or only minor concentrations of, unwanted contaminants appear; optical resolution steps are not necessary (as in the case of most chemical-processes) since only the L-form is synthesised; the required amino acid can be relatively easily separated from cells and protein impurities. By making use of the physicochemical properties of the amino acid that is required, it is possible to obtain highly purified and/or concentrated product. This is done by a combination of several processes, the number of which is dictated by the final application for the product and by economical feasibility. Specific methods to separate the amino acid required from its contaminant products, such as medium components and other amino acids, are plpceeded by removal of the cells and proteineous material from the culture broth. Following cell and proteineous material removal, one or more specific methods of amino acid recovery are applied. These are: aystallisation; 0 ionexchange; 0 electrodialysis; 0 solvent extraction; 0 decolorisation; 0 evaporation. We will now consider each of these methods in turn. rnelhodsbr amino acid r-ery 8.5.1 Separation of cells and proteins from the fermentation broth Two methods are commonly used, these are centrifugation and filtration. Both methods are used on a large scale and are economically feasible. Centrifugation can be operated on a semi-continuous or continuous basis and there are several different types of centrifuges. Large scale tests have to be performed to choose the proper centrifuge (unloading speed, capacity, separation performance etch centrifugation Sometimes poor centrifugation behaviour of cells can be improved by adding flocculation agents. These agents neutralise the anionic charges (carboxyl and phosphate groups) on the surface of the microbial cells. Examples of flocculation agents are alum, calcium and ferric salts, tannic acid etc. 250 Chapter 8 fihhll filleraids PH Oemp8fatUEl Salts cation and anion exchange resins Filtration is another possibility to remove the cells from the amino acid containing broth. Factors to be considered are: 0 properties of the filtrate (solid/liquid ratio); 0 nature of the solid particles (type of micr~rgani~m); 0 adequate pressure (or vacuum) to obtain adequate flow rates; negative effects of antifoaming agents on filtration. Sometimes filtration can be improved by using filteraids. These filteraids, which are based on diatomaceous earth, improve the porosity of a resulting filter cake leading to a faster flow rate. Before filtration a thin layer is used as a precoat of the filter (normally standard filters). After that a mix is made with the harvest broth and filtration is started. 8.5.2 Crystallisation Crystallisation is often used as a method to recover the amino acid. Because of the amphoteric character (contains both acidic and basic groups) of amino acids, their solubility is greatly influenced by the pH of the solution and usually show minima at the isoelectric point (zero net charge). Since temperature also influences the solubility of amino acids and their salts, lowering the temperature can be used in advance as a means of obtaining the requved product. Precipitation of amino acids with salts, like ammonium and calcium salts, and with metals like zinc are also commonly used. This is followed by acid (or alkali) treatment to obtain the free or acid form of the amino acid. 8.5.3 Ion exchange Ion exchange resins have been widely used for the extraction and purification of amino acids from the fermentation broth. The adsorption of amino acids by ion exchange resins is strongly affected by the pH of the solution and by the presence of contaminant ions. There are two types of ion exchange resins; cation exchange resins and anion exchange resins. Cation exchange resins bind positively charged amino acids (this is in the situation where the pH of the solution is lower then the isoelectric point (IEP) of the amino acid), whereas anion exchange resins bind negatively charged amino acids (pH of the solution is higher than IEP). Elution of the bound amino acid(s) is done by introducing a solution containing the counterion of the resin. Anion exchange resins are generally lower in their exchange capacity and durability than cation exchange resins and are seldom used for industrial separation. In general, ion exchange as a tool for separation is only used when other steps fail, because of its tedious operation, small capacity and high costs. Examples given in literature are: 1) strongly basic anion exchange (OH) for the separation of 0 L-glutamic acid from broth; Industrial production of amino acids by fermentation and chemo-enzymatic methods 251 net zero charge only small dstribulion coefficients activated charcoal 0 basic amino acids from neutral and acidic amino acids; 0 separation of aromatic amino acids (phenylalanine and tyrosine) from other amino acids. 2) strongly acidic cation exchange (H+; IWG) for the separation of 0 L-proline adsorbtion and elution with diluted HCl; 0 L-threonine adsorbtion and elution with aquous ammonia. 8.5.4 Electrodialysis This method is based on the principle that charged particles move towards electrodes in an electric field. A mixture of the required amino acid and contaminant salts can be separated at a pH where the amino acid has a net zero charge (at the IEP). The salt ions are captured by the ion exchange membranes that are present. The applications are limited to desalting amino acid solutions, eg removal of HCI from L-glutamic acid solution. 8.5.5 Solvent extraction This method also has only limited applications. The reason for this is that the distribution coefficients of amino acids between organic solvent and water phases are generally small. There are some possibilities given in literature which are based on the alteration of the amino acid. Examples are: 0 cyclisation of L-glutamic acid and extraction with alkyl and aromatic alcohols; 0 conversion of contaminant organic acids (like acetic acid) to the ester form and extraction of the ester; 0 extraction of basic amino acids (like L-lysine) from aqueous solution with water immiscible solvents containing higher fatty acids; 0 conversion of L-isoleucine/L-leucine mixtures to their cobalt salts and subsequent extraction of L-isoleucine by low molecular weight alcohols; 0 conversion of amino acids to their aldehyde derivative and subsequent extraction etc. 8.5.6 Decolorisation This step is almost always performed to get rid of the coloured impurities in the fermentation broth. The method is based on the fact that amino acids (especially the non-aromatic amino acids) do not adsorb onto activated charcoal. Although the treatment is very effective, some of the amino acid is lost during this step. Alternative ways are: use of porous ion exchange; 252 Chapter 8 addition of cationic surfactants, high molecular synthetic coagulants or some phenolic compounds; 0 washing of crystals with weakly alkaline water as in the case of glutamic acid. 8.5.7 Evaporation Evaporation of the amino acid containing solution is a quick but commercially unattractive way (high energy costs) to obtain amino acids from solution. This can only be used when the total amount of contaminant products is very low, since these compounds are not removed and appear in a concentrated form in the product. rI of amino acids from fermentation broth. qudcbut eVm*e Make a list of the main merits and limitations of the various methods of recovery Method Merit Limttation Centrifugation Filtration Crystallisation Ion exchange Electrodialysis Solvent extraction Decolorisation Evaporation In the recovery of amino acids from fermentation broths, several methods can be applied. Match each of the following statements to one or more separation methods. Statement 1) Separation efficiency is related to a distribution coefficient. 2) Separation relies on the amphoteric natum of amino acids. 3) Separation may be improved by addition of calcium salts. 4) Separation involves adsorption onto activated charcoal. 5) Separation method can only be used if total amounts of contaminating products is very low. Industrial production of amino acids by fermentation and chemo-enzymatic methods 253 8.6 Case study: fermentative production of L-phenylalanine from glucose In this section we will examine the fermentative production of L-phenylalanine in detail. This will serve to illustrate some of the general principles of fermentation already considered. Phenylalanine production from a plasmid harbouring auxotrophic mutant of E. coli has been studied in batch cultures. Phenylalanine is produced in the stationary phase due to the release of feedback inhibition after depletion of tyrosine. The fermentation can be summarised as: summary reaction CH, - CH- COOH + byproducts cells C,H120, biomass + glucose N¡±2 L-phenylalanine acetic acid, C02 The characteristics of the strain are: Atrp, Atyr /pJN6 carrying the genes amF, pheAFR, TcR and ApR. Explanation: A = auxotrophic for tryptophan and tyrosine; FR = feedback resistant to phenylalanine; R = resistant to the antibiotics tetracycline and ampicillin. The composition of the basic medium is (s 1-l): medium composition NH4CI: wo4: KHzPO4: NeHP04: Na-citrate: CaCI2.2H20: tyrosine: tryptophan: tetracycline: glucose: FeC13.6H20: MgCI2.6HD: 5.0 0.8 0.5 1 .o 2.5 0.01 0.02 0.8 0.05 0.025 0.01 0 35 Nitrogen source Samponent Pcomponentlbuffer Pamponentlbuffer Carbon source trace elements: 1 ml I¡± 254 Chapter8 These elements are necessary for maintenance and growth of the organism and for product synthesis. The pH of the medium is set around 7.0 (neutral conditions) and sterilisation is carried out by autoclaving at 121°C. CaC12.2H20, MgCb6HD and glucose are sterilised separately. In the first two cases to prevent precipitation and in the last case to prevent caramelisation (decomposition of glucose). Tyrosine, tryptophan and tetracyline are filter-sterilised to prevent decomposition by heat sterilisa tion. Three 1 litre baffled flasks, each containing 100 ml medium, are inoculated with cells from one agar plate suspended in 10 ml saline and incubated at 30°C on a rotary shaker (for optimum supply of oxygen). This provides sufficient biomass to inoculate the bioreactor. swmation inoculation A 3 litre bioreactor with a working volume of 2 litre is inoculated with the three shaking flasks. The pH is maintained at 5.5 by automatic titration with 5mol1-' WOH and the temperature is held at 3yC. The agitation speed is set a 600 rev min-' and the air flow rate at 1.51.6 litre min-'. The following parameters are measured more or less continuously: 0 dissolved oxygen tension, using an in situ oxygen electrode; phenylalanine, by test kit; monitoring he fermentation 0 glucose concentration, enzymatically; 0 the byproduct acetic acid, by gas chromatography; biomass, by optical density and dry weight; temperature, using temperature probe; Co2, by infrared gas analysing system. Measurement of all these parameters provides sufficient information to evaluate the fermentation and leads to the economical production of the amino acid. We will now discuss several relevant parameters seperately with the help of the data obtained (see Figure 85). Industrial production of amino acids by fermentation and chemo-enzymatic methods 255 Figure 8.5 The production of phenylalanine from E. coli. Biomass yield Biomass production is determined by measuring the optical density at 620 nm and/or weighing after drying to constant weight. The typical form of the growth curve is given in Figure 8.5 (open triangles). We can recognise at least three stages: The lag phase (circa 2 hours) during which the cells are not capable of multiplying; The exponential (log) phase (up to 8-10 hours) where cells multiply most rapidly; . The stationary phase where growth ceases (10-25 hours); The death phase where some cells are lysed (after 25 hours; not obvious in Fip 85). We can also see from Figure 8.5 that glucose is consumed rapidly during growth (open squares) and that almost all glucose is consumed by the end of the fermentation. Phenylalanine production and byproducts Production of phenylalanine starts after depletion of tyrosine at about 6 hours. This is logical since the miao-organism needs a certain amount of tyrosine, for example to synthesise key enzymes, but synthesis of L-phenylalanine is feedback rrgulated if tyrosine is present. stages on batch 9rd cyde depletion of WO~~ 256 Chapter 8 Lphenylalanh production is largely independent of the growth stage (see Figure 85, closed circles), although maximum specific rate of phenylalanine production (ie the amount of Gphenylalanine produced per unit of time) is at the end log phase of the growth (open circles). For economic reasons it is worthwhile to know the amount of L-phenylalanine that is produced per amount of raw material (glucose, carbon source). This can be calculated from Figure 85 after 14 hours during constant product formation and is called yield of L-phenylalanine from glucose. As in most fermentations, byproducts are formed: part of the carbon flow is dkcted to the tricarboxylic acid cycle and acetate is formed in low amounts. product yield bmgb- Carbon balance As we already know, glucose is the major carbon input. Carbon from tyrosine, tryptophane and citric acid is negligible and can be excluded from the calculation. Carbon from the substrate glucose is converted into the carbon of the cells, phenylalanine, carbon dioxide and byproducts. Carbon balance calculations thus give us more understanding of the amount of carbon in glucose used for cell mass production, for synthesis of the wanted product, maintenance energy and byproduct formation. In the stationarv phase, assuming that the total cell mass in the reactor is constant and readon that only L-phiylalakine and-acetate are produced, the following stoichiometric sDJichiome~ equations are used: 26H1206 + NH3 + 202 -+ C9Hl10rN + 8H2O + 3C02 (glucose) (phenylalanine) (Equation 8.1) C&~a + 202 + 2cIGCOOH + WZO + 2C02 (glucose) (acetate) (Equation 82) From the literature the following stoichiometric equation for E. coli growing aerobic on glucose is well known. C&&j + O.lNH3 + 8.45702 -+ O.~CH~~O.~~NO~~ + 5.5C02 + 5.7 HD (Equation 8.3) Using these equations it is possible to calculate theoretical values for glucose yield coefficients (YG). YC values can be calculated for cell mass, phenylalanine and acetic acid and is simply the amount of product produced per unit amount of glucose consumed. So the theoretical yield of phenylalanine on glucose, for example using equation 8.1, is: 0.5 mol phenylalanine (mol glucose)-' or 0.47 g phenylalanine (g glucose)-'. Of course this is an absolute maximum value since part of the glucose that is supplied is used for byproducts such as acetate. In such a case it is more realistic to combine both equations (8.1) and (8.2): 0.33 mol phenylalanine (mol glucose)-' or 0.31 g phenylalanine (g glucose)-'. gh~coseyield coefficient Industrial produdion of amino acids by fermentation and chemo-enzymatic methods 257 expedimentally derived data Now consider the experimental results shown in Figure 85. From these graphs data can be extracted and presented in tabular fom. Table 8.4 presents some data extracted from Figure 85, along with other data (for carbon dioxide) that were not shown in the Figure. Tlme(h) 2 4 6 8 10 14 24 32 48 glucose 34 33 30 28 26 18 15 12 4 (a r¡¯) dry 0.3 0.6 1.7 2.0 2.2 2.3 2.3 2.3 2.2 (9r 1 (9 r¡¯) (g 1-7 weiqht phenyla- 0.0 0.02 0.1 0.3 0.6 0.8 1.6 2.2 3.0 lanine acetate 0.1 0.2 0.3 0.2 0.2 0.0 0.1 0.3 0.8 COP 0.25 0.5 0.6 0.7 0.7 0.6 outlet culture 2 2 2 2 2 2 2 2 2 volume Table 8.4 Data obtained from Figure 8.5. Use the data given in Table 8.4 to calculate the experimental yield of phenylalanine on glucose [g phenylalanine <s glucose)-¡¯] for the interval between 14 and 32 hours: n yieldof Ph@Mhdm For the interval between 14 and 32 hours: glucose consumed = 18-12 = 6 g 1-I; phenylalanine produced = 2.2-0.8 = 1.4 1-¡¯. So, yield of phenylalanine on glucose = 1.4/6 = 0.23 g phenylalanine (g glucose)- . The maximum yield, as can be seen from Figure 85, is in the late exponential phase. Using the data given in Table 8.4 for the interval between 14 and 24 hours, we obtain a value of 0.27 g phenylalanine (g glucose)-¡¯ for the maximum yield coefficient. Now compare the maximum experimental YG value for phenylalanine (0.27 g g-¡¯) n with the theoretical YG value (0.31 gg¡±). How can we account for this discrepancy? This discrepancy is due to the fact that other products such as formate, are formed in very small amounts as byproducts of the metabolic routes leading to L-phenylalanine and polymer synthesis. Of course, part of the glucose is also used for the metabolic activities in the microorganism necessary to maintain the cells in a viable state, this is termed the maintenance energy requirement. To find out where the carbon flow goes to we can simply investigate Figure 8.5. It is obvious that: mainsnance energy requirements carbon now in the early exponential phase most of the glucose is used for biomass production; 258 Chapter 8 in the midexponential phase part of the glucose is used for the production of phenylalanine. This portion increases entering the late exponential phase; in the stationary phase there is no cell growth, so most of the glucose is used for the production of L-phenylalanine. consumption, in moles, in the interval between 14 and 32 hours ?lucose relatwe molecular mass of glucose = 180). 2) Now calculate the number of moles of glucose necessary to maintain the reactions in this interval. What can you deduce from these calculations? Hint Consider the number of moles of phenylalanine and acetic acid formed and the stoichiometric equations 8.1 and 8.2. (RMM phenylalanine = 165, acetic acid = 60). 1) From the data given in Table 8.4 calculate the observed 8.6.1 Process economics of the fermentative production of L-phenylalanine The ultimate goal of process development is to achieve feasibility: where it is possible to produce amino acids on a large scale at a production cost per kg of amino acid comparable to, or cheaper than, the processes currently used by other companies. If we presume that the technical performance (fermentation and recovery) are sorted out on a laboratory scale and scaling up looks promising, then it is time to find out whether it is possible to operate economically on a large scale. The thing to do first is a cost price evaluation on the particular process. It is obvious that one should know as much as one can of the process in order to come to a realistic cost price evaluation. Since these data are not always available in the literature (especially not for ongoing commercial fermentation processes) we have to make use of those general data that are available in the literature and patents. First we start with a general prescription of the process. The aim, for example, is to make 1,000 tonnes of L-phenylalanine a year (1 million kg per annum). Presume that a yield of 20 g 1-' of L-phenylalanine can be obtained. This is more realistic, based on patent literature, than the low yields in the example considered previously (section 8.6). This automaticall means that more glucose will be needed. Let us again presume that instead of 35 g 1- glucose we now need 150 g 1-' to achieve this overall yield. The concentrations of the mineral salt are kept the same to maintain good buffering capacity, whilst the concentration of tyrosine and ~tophan are also increased by a factor 4.3 (tyrosine 0.21 g 1-' and tryptophan 0.11 g 1 ) A reasonable size of bioreactor, based on transport and handling considerations, is 200 m3, with a working volume of 150 m3. If the fermentation time is 48 hours and down time for reuse about 24 hours, then the total batch time is 72 hours. As a rule of thumb, total investment of the fermentation plant is 50 dollars per litre of bioreactor. This means the total costs of investment are 200,000 x 50$ = 10 million dollars. This figure includes all costs concerning engineering, fermentation equipment, recovery equipment, buildings land, etc. cost Price evalualion Y size of bi-ct~r Industrial production of amino acids by fermentation and chemo-enzymatic methods 259 The second calculation is on the production costs themselves. We can discriminate three majorparts: 0 variable costs; total direct costs; plant gate costs. Variable costs are costs for raw material used in the fermentation and for recovery (for example if flocculants are used) and utilities like steam (sterilisation, water evaporation), water (medium make up, washing, cleaning, cooling etd electric power etc. variable costs Let us now reconsider the medium described previously in section 8.6 but adapted for an industrial fermentation. Mapr costs for raw materials are those for glucose, tyrosine, tryptophan and salts. The figures are given in Table 85. kg kg -' phenylalanine price $ kg-' $ kg-' phenylalanine glucose 7.5 0.40 3 salts tyrosine tryptophan 0.53 0.01 1 0.006 1.13 30 100 0.6 0.32 0.6 I Total 4.52 $ kg-' Table 8.5 Raw materials costs for industrial scale fermentation of L-phenylalanine. As mentioned previously, utilities are steam, water, electricity and gas necessary to run the plant. It is extremely difficult to obtain proper figures from the literature. In addition, costs of utilities differ from plant to plant and from country to country. Not knowing how much of each of these utilities are necessary to run this fictitious plant we have to use "fake" figures. Based on information in the literature, th~ costs for utilities are about 10% of the total variable costs, in our case being 0.50 $ kg- phenylalanine produced. Variable costs are summarised in Table 8.6. Direct costs are costs for maintenance of materials, operating of supplies labour, control If the minimal number of operators necessary to run the plant is for example 10 in a 4 shift system all the year through, than there is a total of 40 operators on the paymll. At an average salary of $20,000 a year, the total dim3 costs for labour are $800,000. Indirect labour accounts for 40% of the direct labour, being $320,000, and 85% of the direct labour is used for salaried payroll ($68O,OOO). Associated payroll costs are assumed to be about 30% of the total payroll costs (indirect+direct+salaried payroll), ie $54om. direct costs laboratory etc. 260 Chapter 8 Maintenance salaries and costs for laboratories are estimated as 3% of the total capital of investment, ie 3% of 30 million dollars is $9OO,OOO. The figures for payroll charges calculated back to one kg of L-phenylalanine are given in Table 8.4. INVESTMENT ($, million) Total fixed capital 30 PRODUCTION COSTS ($ kg-¡¯ L-phenylalanine) Variable costs raw materials 4.52 utilities 0.50 total variable costs 5.02 Dlrect costs direct labour indirect labour salaried payroll associated payroll maintenance salaries, laboratories etc Supplies and expenses: maintenance supplies total direct costs 0.80 0.32 0.68 0.54 0.90 0.90 0.20 4.34 Plant gate costs depreciation (15 years) taxes and insurance plant overhead 2.00 1.20 0.48 total plant gate costs 3.68 total production costs (fixed + variable are thus $13.04 kg-¡¯) plant gab Costs Table 8.6 Cost price evaluation of L-phenylalanine fermentation based on a production of 1,000 tonnes. As part of the total direct costs we also have to consider costs made for supplies and expenses: maintenance, 3% of total capital; supplies, 25% of dired labour. These figures, calculated back to one kg of L-phenylalanine are given in Table 8.6. Plant gate costs are other costs concerning the fermentation plant; for example depreciation, taxes and insurance and other plant overhead costs. Fifteen years depreciation is not abnormal. This brings the depreciation costs to $2 million a year. Taxes and insurance are estimated to be 4% of the total capital being $1.2 million a year. Plant overhead accounts for about 60% of total direct labour and is thus calculated to be !§48O,OCHl (see Table 8.6). Industrial production of amino acids by fermentation and chemo-enzymatic methods 261 return on inv-mt aitefion Now that we have dealt with most cost factors, lets see whether the process is profitable or not. A key factor in industry is profitability regardless of technical achievement. Objective procedws to aid such assessment are based on the return on investment (ROI) as the criterion. This is calculated from the following assumption: ROI = (A + 9B)¡¯ lo x 100 (expressed as a percentage) C+D (Equation 8.4) where: ROI = return on investment; A = net earnings after taxes (50%) in the first year; B = net earnings after taxes (50%) per annum for the coming period in which the investment should be payed back; C = on@ fixed capital investment; D = working capital (25% of net sales). To calculate the ROI, again we have to make some more assumptions. Let us assume that we would like to have our costs of investment back in say 10 years, so we can start making real money after that. Let us in addition assume that the selling price for L-phenylalanine is 30 to 40 $ kg-¡¯. At a production of LOO0 tonnes, this means a total sales of $30 to $40 million a year. After extraction of costs such as dealer discount, distribution and freight (total about 30%) the net sales are $21 to 28 million a year. Since we calculated a cost price of about 13 dollars per kg phenylalanine (see Table 8.6; $13 million per 1,OOO tonnes), the gross profit will be 8 to 15 million dollars. Assume 12.5% of the net sales for cost of sales and adminstration plus research and development (ie 2.6 to 3.5 million dollars) the profit from operations in the coming period will be 5.4 to 11.5 million dollars. In the first year, the startup costs are substantial (10% of the capital; in our case 3 million dollars). The profit for the first year is, therefore, not 5.4 to 11.5 million dollars but 2.4 to 85 million dollars. 10 Year pwCbion The net earnings after taxes (5096) in the first year are then $1.2 to $4.25 million (A in equation 8.4) and the net earnings per annum for the period to come are $2.7-$5.75 million (B in equation 8.4). The working capital is normally about 25% of the net sales ($5.25 to $7 million; D in equation 8.4) whilst the total fixed capital was assumed to be $30 million (see Table 8.6; C in equation 8.4). The ROI calculated from equation 8.4 will then be 7 to 15%. Usual figures for ROI in the literature are 15 to 20%. The total production costs will be increased with sales and admipstration plus research and development, so the final production costs will be 13.04$ kg- + ($2.6 to $350) giving a total production cost at 100% capacity of $15.64 to $16.54 per kg of phenylalanine. 262 Chapter 8 The product value at 100% capacity will now be (total cost of production + 7 to 15% ROD, ie $16.04 to $16.54 + $1.12 to $2.48. So the minimum product value will be $17.16 per kg of L-phenylalanine and the maximum product value $19.02 per kg of Gphenylalanine. It is rather difficult to say whether this fictitious process would survive or could compete. Actual data are absolutely necessary. On the other hand this exercise gives us a better understanding of p'ocess economics and can also be used to compare a fermentative process for the production of amino acids with, for example, a chemo+nzymatic process. Calculate the retum on investment over a 15 year period for an amino acid fermentation, based on the following data and assumptions. Production capacity = 500 tonnesper annum Selling price of product = 50 $ kg- Cost price of product = 24.5 $ kg-' Capital = $40 million Taxes = 50%. ASSUmptiOnS Cost of dealer discount, distribution and freight = 20% total sales Startup costs = 10% of capital Working capital = 25% of net sales Administration plus R and D costs = 12.5% of net sales. 8.7 Case stud : The production of L-phenylalanine by enzymatic me r hods Before we deal with the use of enzymes for the production of amino acids, we suggest that you may wish to refresh your memory of the material covered in Chapter 2. To illustrate a few aspects of amino acid production by enzymatic methods, the production of Lphenylalanine will be considered in some detail. L-Phenylalanine is important as an essential amino acid for human nutrition and is used as an intermediate for the synthesis of the artificial sweetener, aspartame. Other examples of the industrial production of amino acids by enzymatic methods are described briefly in Appendix 2. In section 8.6 we saw that L-phenylalanine can be produced from glucose by fermentation. In this production method mutants are required to achieve high yields of L-phenylalanine. Reaction schemes for the production of L-phenylalanine by enzymatic methods are shown in Figure 8.6. Although the term bioconversion can be used to describe all types of productions discussed in this section, it generally refers to the process conducted by micro-organisms and/or enzymes with unnatural substrates (which are not present in normal biochemical pathways). The enzymes available in or from the biomass are used for conversion of the unnatural substrate to obtain the desired product. Two of the main raw materials used for bioconversion to L-phenylalanine are trm-cinnamic acid and acetamido cinnamic acid (reactions 1 and 2 in Fip 8.6.) umahrral subsmates Industrial production of amino acids by fermentation and chemoenzymatic methods 263 Figure 8.6 Reaction schemes for the production of L-phenylalanine by enzymatic methods. 264 Chapter 8 Figure 8.6 Continued. Reaction 1 is governed by the enzyme phenylalanine ammonia lyase. This enzyme normally conducts the breakdown of L-phenylalanine to trans-cinnamic acid and ammonia. However, the reaction can be reversed leading to the production of L-phenylalanine from -Cinnamic acid by using excess ammonia. Much research deals with the production of L-phenylalanine from acetamidocinnamic acid (ACA) because of the low price of ACA. Different pathways have been reported but the principles of the main reaction pattern are given in reaction number 2 (Figure 8.6). At least two enzymes, an acylase and an amino transferase, are necessary for the bioconversion of ACA to L-phenylalanine. The amino source usually is an amino acid, Laspartic acid is often used. The production of L-phenylalanine by precursor addition is given in reaction 3 (Figure 8.6). The use of intermediates as substrates in Lphenylalanine synthesis avoids inhibition by metabolites. Phenylpyruvic acid, an intermediate precursor in the biosynthesis of L-phenylalanine, can be converted to L-phenylalanine. Gaspartic acid is often used as an amino donor. The amino group can only be transferred from an rmnscimmk acicl acetamid0 dnnmicacid prearrsar addmin Industrial produdion of amino acids by fermentation and chemo-enzymatic methods 265 organic nitrogen source and needs a cofactor (pyridoxalphosphate), which is often present in whole cells. So ammonia cannot be used for amino group addition. The optical resolution of the chemically synthesised N-acetyl-DL-phenylalanine by an acylase enzyme is given in reaction 4 (Figure 8.6). A selective hydrolysis of N-acetyl-L-phenylalanine is performed. Two reactions for the production of L-phenylalanine that can be performed particularly well in an enzyme membrane reactor (EMR) are shown in reaction 5 and 6. The recently discovered enzyme phenylalanine dehydrogenase plays an important role. As can be seen, the reactions are coenzyme dependent and the production of L-phenylalanine is by reductive amination of phenylpyruvic acid. Electrons can be transported from formic acid to phenylpyruvic acid so that two substrates have to be used: formic acid and an a-keto acid phenylpyruvic acid (reaction 5). Also electrons can be transported from an a-hydroxy acid to form phenylpyruvic acid which can be aminated so that only one substrate has to be used: a-hydroxy acid phenyllactic acid (reaction 6). We can summarise the enzymatic methods of L-phenylalanine production as follows: 0 bioconversion (trans-cinnamic acid or acetamidocinnamk acid as bioconversion substrate); 0 precursor addition (phenylalanine as precursor); 0 optical resolution (N-acetyl-DL-phenylalanine addition); 0 reductive deamination involving coenzyme regeneration (a-keto acids or a-hydroxy acids addition). Consider reaction schemes for the production of L-phenylalanine by enzymatic methods. Now match each of the following substrates with the enzyme(s1 responsible for L-phenylalanine formation. Substrates -t optical resoluhn re,j,,dve diaminatim 1) Phenylpyruvic acid (as precursor) 2) N-acetyl-DL-phenylalanine 3) Acetamidocinnamic acid 4) Phenylpyruvic acid (reductive deamination of a-keto acid) 5) trans-cinnamicacid 6) Phenyllactic acid (a-hydroxy acid). Enzymes L-hydroxy acid dehydrogenase acylase phenylalanine-ammonia-1 yase Lamino acid dehydrogenase L-amino acid amino transferase 266 Chapter 8 bansamhation of phenylpyruvic acid fed-batch fermentation Ol[vpl limrtabon resting cells product fonnation Now that we have seen the possible enzymatic routes to overproduction of L-phenylalanine, in the next section we will consider two p")cesses for production of this amino acid in more detail. These are precursor addition using phen lpyruvic acid processes are carried out using whole cells in a kind of semi-fermentation. We will see that, in comparison with direct fermentation using glucose (section 8.61, productivity for the semi-fermentations are good although economic feasibility is strongly influenced by the additional cost of the substrate. 8.7.1 Production of L-phenylalanine by precursor addition and bioconversion using acetamidocinnamic acid. These essentia Y ly enzymatic The production of L-phenylalanine from the precursor phenylpyruvic acid by transamination is a process which requires two steps: 1) cell production; 2) L-phenylalanine production after addition of phenylpyruvic acid. Cell production can be camed out by a normal fed-batch type of fermentation. The feed rate of glucose is increased during the fermentation and the cells grow exponentially. Biomass formation and transamination activity within the cells develop in a similar manner. Growth usually continues until limited by the availability of dissolved oxy^ tension (DOT). After 10-15 hours a dry weight biomass concentration of 10 g 1- is normally reached. The culture can be used directly for the conversion of phenylpyruvic acid to L-phenylalanine. Therefore, a batch process with resting cells can be camed out, with some glucose added for maintenance (fed-batch fermentation). Another approach is to harvest the cells from the fermentation broth and to use them in a separate bioreactor in higher concentrations than the ones obtained in the cell cultivation. An advantage of the last method can be that the concentration of compounds other than L-phenylalanine is lower, so that downstream processing may be cheaper. Figure 8.7 gives a typical time course for the conversion of pyruvic acid to L-phenylalanine. Experiments for optimisation of the production of L-phenylalanine are usually camed out at pH = 75, a temperature of 3TC, 50% DOT and 10 gram dry weight biomass per litre medium. Maximum productivity is reported to vary between 3 and 6 g 1-' h-' and product concentrations of between 11-28 g 1" have been mported. The time necessary for completion of the reaction is about 8 hours (see Figure 8.7). Industrial production of amino acids by fermentation and chemo-enzymatic methods 267 Figure 8.7 Fed-batch fermentation of phenylpyruvic acid to L-phenylalanine. Specific information about the optimum conditions for the synthesis and the activity of the enzyme has been reported for Pseudomonas fZuorescens: screening of various micro-mganisms resulted in the selection of a P. fluorescens strain with an initial rate of conversion of 3 g 1-' h-' in an unoptimised state. The following conclusions muld be made concerning the production of L-phenylalanine by P. fluorescens: The activity was generally higher in cells grown in a cultivation medium containing D- or L-phenylalanine, this confirmed the inducible nature of the phenylalanine aminotransferase. The elimination of the amino donor, L-aspartic acid, resulted in an almost complete reduction of activity. Neither cell permeabilisation nor cofactor (pyridoxalphosphate) addition were essential for L- hen lalanine production. Maximum conversion yield occurred (loo%, 22 g 1-1 when the amino donor concentration was increased. Aspartic acid was a superior amino donor to glutamic acid; 35 g I-' was used. Maximal conversion was observed at 3040°C and at alkaline pH. pH 11 resulted in very rapid initial conversions (12 g I-' h-'1, which decreased to a rate of approximately 3 g I-' h-I. The high conversion rate at high alkaline pH illustrates the potential for biochemical manipulation of this strain. The solubility of L-phenylalanine is greatly increased at such high pHs and enzymes catalysing the degradation of L-phenylalanine and the formation of byproducts may be significantly inhibited at such high pHs. Furthermore, at this high pH the phenylpyruvic acid substrate will rapidly equilibrate between the keto and enol forms, so facilitating the future use of high concentrations of substrate with little inhibition of the reaction rate by the keto acid itself. The final conversion yield decreased when substrate concentration was increased from 2% to 4%. This was attributed to end product inhibition by the L-phenylalanine produced. Thus although faster conversion rates were observed with addition of high substrate concentrations, the product titres never exceeded 16 g 1". As already discussed the rate of yield of the conversion was proporti~~l to the concentration of amino donor employed. Using a ratio of 1:3 substrate to amino donor, almost a 90% conversion was achieved in 3 hours. constitutive amino- transferase influence of PY amino donor influence of pH end product inhibition 268 Chapter 8 The optimum cell concentration was between 10 and 20 g 1-', which is not different from the concentrations employed in direct fermentation. A rapid activity loss (a few days) was observed with whole cells. Immobilisation increased the stability and continuous production of L-phenylalanine was possible using alginate bead immobilised cells of P. fluurtscens for 60 days. However, to achieve this the cofactor pyridoxalphosphate had to be continuously added to the beads to correct for the dissociation of the cofactor and loss from the cells. It would seem that immobilisation looks very promising, but remember that the costs of immobilisation and the addition of cofactor have to be compared with the cost of production by free cells. The results obtained with the P. fluurescens strain without biochemical manipulation compare well with those reported for a E. coli strain. Both achieve volumetric rates of 3 g 1-' h-' under nod conditions. So it appears that the efficiency of the process can be increased by a few simple operations: increasing the pH and amino donor concentration (aspartic acid). A possible explanation for the superiority of the amino donor, L-aspartic acid, has come from studies camed out on mutants of E. coli, in which only one of the three transaminases that are found in E. mli are present. It is believed that a branched chain transaminase, an aromatic amino acid transaminase and an aspartate phenylalanine aspartase can be present in E. coli. The reaction of each of these mutants with different amino donors gave results which indicated that branched chain transminase and aromatic amino acid transminase containing mutants were not able to proceed to high levels of conversion of phenylpyruvic acid to L-phenylalanine. However, aspartate phenylalanine transaminase containing mutants were able to yield 98% conversion on 100 mol 1-' phenylpyruvic acid. The explanation for this is probably that both branched chain transaminase and aromatic amino acid transminase are feedback inhibited by L-phenylalanine, whereas aspartate phenylalanine transaminase is not inhibited by L-phenylalanine. In addition, since oxaloacetate, which is produced when aspartic acid is used as the amino donor, is readily converted to pyruvic acid, no feedback inhibition involving oxaloacetate occurs. The reason for low conversion yield of some E. coli strains might be that these E. coli strains are deficient in the aspartate phenylalanine transaminase. Application of the desired biotransformation to give a practical and economical process required high molar conversion yields, high amino transferase activities, highly efficient product recovery and an inexpensive source of phenylpyruvic acid. With genetic and/or biochemical manipulation considerable progress can be made towards meeting some of these requirements. Consider the production of L-phenylalanine using P. fluumcms by precursor (phenylpyruvic acid) addition. Explain briefly why: dl concentralion cell km*ihtjon muantstrains Of E cdi 8spaFw phenylalanne transaminase 1) Growth of cells in presence of D- or L-phenylalanine is desirable. 2) An alkaline pH is preferred. 3) Precursor concentration in the medium should be less than 4%. 4) Cell immobilisation is costly in comparison to the production of free cells. Industrial production of amino acids by fermentation and chemo-enzymatic methods 269 min0tranS- ferase hfluence of pH amino donor cell concentraltions 8.7.2 Bioconversion of acetamidocinnamic acid Recent progress published on the production of L-phenylalanine by bioconversion of acetamidocinnamic acid (ACA) martions selection of s@al highly productive strains. Organisms were obtained by screening for growth on ACA as C and N source (which I.equires acylase activity) and testing for L-phenylalanine accumulation. The enzymatic activity for forming L-phenylalanine from ACA was considered to be inducible since this activity was not detected when ACA was omitted. It was proven that the pathway of L-phenylalanine formation involved phenylpyruvic acid as intermediate and two steps could be distinguished (see Figure 8.6; section 8.7): 1) an ¡®acylase¡¯ enzyme that splits the acetyl group and gives rise to phenylpyruvic acid; 2) a reaction catalysed by a pyridoxal-dependent enzyme that binds the amino group and gives rise to L-phenylalanine formation (transaminase reaction). Two strains producing high L-phenylalanine levels were chosen for experiments: Alcaligenes faecalis and Bacillus sphaericus. Both strains showed equal production rates of 3.3 g 1-I d-¡¯ at WC, pH 7 and 40 g 1-¡¯ biomass. It was observed, however, that it was difficult to elevate aminotransferase activity together with acylase activity. Further research was therefore conducted on two special strains, each performing a specific reaction step in the formation of L-phenylalanine from ACA. One strain with a high acylase activity and the other with a high aminotransferase activity. These strains were obtained by screening for aminotransferase and acylase activity, resulting in B. sphericus as an acylase specialist and Pmucoccus d&fim as an aminotransferase specialist. Acylase activity in the cells of B. sphaericus was only induced in a medium with acetamidocinnamk acid. To establish the most advantageous conditions for production of L-phenylalanine from acetamidocinnamic acid using two micro-organisms the following factors were investigated: pH, amino donor and ratio of two enzyme activities. The optimum pH for both enzymes was 7.5-8.0. For practical production it is desirable to carry out a reaction in a single reactor at optimum pH. Therefore, it is most advantageous to carry our the two step enzyme reaction at pH 8.0. The best results were obtained with L-aspartic acid as the amino donor for P. denitrifiuzns and phenylpyruvic acid as the amino acceptor. With L-aspartic acid, conversion of phenylpyruvic acid exceeded 90%. This may be attributed to absence of feedback inhibition of the reaction due to metabolism of the reaction product, oxaloacetic acid. When using glutamic acid the conversion of phenylpyruvic acid did not exceed 60%. Practical conversion from ACA to L-phenylalanine was best achieved when acylase and aminotransferase activities were equal. This could be achieved by using 10 gram B. sphaericus and 25 gram P. denitri*fum per litre medium. Under optimal conditions (pH = 8.0,67 g 1-¡¯ L-aspartic acid, 30°C, 1:l ratio of enzyme activities) after addition of pyridoxal phosphate, 76 g 1-I L-phenylalanine could be produced within 72 hours (92% conversion). This illustrates how simple biochemical manipulation can increase productivity dramatically. 270 Chapter 8 A cell immobilisation method for production of L-phenylalanine from ACA has also been investigated. Co-immobilised cells showed highest activity. This might be explained by the fact that phenylpyruvic acid produced by the acylase reaction was immediately converted to L-phenylalanine by the aminotransferase reaction without diffusion resistance of phenylpyruvic acid. A ratio of 2:l acylase&otransferase was better for a stable L-phenylalanine forming activity as compared to a 1:l ratio. This mightbecausedbythedifferenceof stabilitybetweenthetwoenzymes.Theoperational stability of L-phenylalanine forming activity inmased with decreasing temperature, so the half-life time increased with decreasing temperature. The activity increased with increasing temperature. The productivity at temperatures higher than 30°C decreased because of low stability of Lphenylalanine forming activity. When the space velocity was 0.06 h-' at 30°C and pH 8.0,24 g 1-I L-phenylalanine was produced with a 98% conversion rate from ACA. A volumetric productivity of 15 g 1" h-' was achieved, 17% higher than with whole cells. The half life under these conditions of the L-phenylalanine activity of the co-immobilised cells was calculated to be 14 days. 8.7.3 Summary of process conditions for L-phenylalanine production The most important process conditions for the production of L-phenylalanine by the three methods discussed in this chapter are summarised in Table 8.7. coimmobiri Fern PPA Aca Units maximum 0.6 3.5 1.5 g r1 h-l productivity time 24 8 60 h PH 7 7.5 8 temp 35 35 35 OC biomass 20 10 40 g I-' ~~~~~~ ~ ~ Table 8.7 The most important process condnions for the production of Lphenylalanine by direct fermentation; precursor addition, phenylpyruvic acid (PPA); bioconversion, acetamidocinnamic acid (ACA). As can be seen from Table 8.7 productivity (expressed in g 1-' h-') is highest for precursor addition. The production of L-phenylalanine from phenylpyruvic acid also has the shortest reaction time to obtain high conversions. The pH commonly used is around 75, quite normal for biological processes. Only the enzyme phenylalanine ammonia lyase shows an optimum pH of 1O.The pmss temperature varies between 30 and @C with an average of 35OC. No extxtme temperatures have been reported due to the fact that denaturation occurs at high temperatures. The optimal concentration for cells frequently used is 10-20 g 1-'. However, conversion of ACA is done with high cell mass concentrations in recent studies; possibly to compensate for substrate inhibition and thus to maintain high product concentration. The processes using PPA and ACA need an amino acid as amino donor, usually L-aspartic acid is used. 8.7.4 Cost of production by precursor addition and by bioconversion Extra cost compared to direct fermentation are mainly concerned with addition of PPA (precursor addition) or ACA (bioconversion). Table 8.9 compaxes estimated costs for the PPA and ACA pms based on the data considered previously (section 8.6.1 and 8.6.2). Costs are estimated for production of 100 tomes per year. Industrial production of amino acids by fermentation and chemo-enzymatic methods 271 PPA $,thousands ACA $, thousands Bioreactor 38 (3m3) 88 Personnel 225 225 Cultivation 328 102 Recovery 1000 lo00 Utilities 59 41 Substrate - PPA 1050 - ACA 1087 Substrate - L-ASP 805 - L-ASP 479 Total 3505 3022 Cost price per kg L-phenylalanine 0.035 0.030 Table 8.8 Estimated costs of production of L-phenylalanine by enzymatic methods (1 00 tonnes per year). PPA = phenylpyruvic acid, ACA = acetamidodnnamic add, L-ASP = L-aspartic acid. We can see from Table 8.8 that substrate costs represent about half the total costs of the production. It follows that the best way of reducing costs is to use cheaper substrates. One possibility is to develop your own production processes for the chemicals. It is not possible to directly compare overall cost of fermentation with the PPA and ACA pmcesses because published information on the processes are for different scales of production (1,OOO tonnes per year for direct fermentation and 100 tonnes per year for PPA and ACA processes). However, we can say that cost savings on direct fermentation is possible when higher product yields are obtained. This wiU result in a lower substrate consumption and a smaller xeactor volume, resulting in lower direct costs. Strain improvement is necessary to achieve significantly higher product yields in direct fermentation. Assume that the cost price of L-phenylalanine produced by direct fermentation is 28.5 $ kg-' (100 tonnes per annum capaaty). What percentage reduction in substrate costs are required for 1) precursor feeding and 2) biotransformation to be competitive on a cost price basis with direct fermentation? substrate costs 272 Chapter 8 Summary and objectives In this chapter we have seen that amino acids can be produced as purr enantiomers by fermentation and by chemosnzymatic processes. In direct fermentation of amino acids, mutant strains or DNA-recombinant micro-organisms are used to overcome feedback inhibition and repression by endproduds of metabolic pathways. Fermentation problems include those associated with back mutation of the process micrmrganism, loss of genetic material and phage contamination. The problems can be minimised by altering the fermentation conditions. Several methods can be used to recover amino acids from fermentation broths. Production of L-phenylalanine by direct fermentation was considered in detail and we saw that carbon balance analysis can be used to characterise the fermentation. Economic evaluation of a bioprocess involves a consideration of investment costs, direct costs, variable, direct and plant gate costs. A key factor in industry is profitability regardless of technical achievement, which is based on the return on investment criterion. L-Phenylalanine can also be produced by various enzymatic processes, including precursor addition (phenylpyruvic acid) and bioconversion (acetamidocinnamic acid). Productivities of these processes are higher than that of direct fermentation. However, precursor addition and bioconversion processes have substantially greater costs than direct fermentation. Other factors that can substantially incmase the cost of chemo-enzymatic processes include the possible requirement for immobilisation of the biocatalyst and for addition of cofactors. Now that you have complete this chapter you should be able to: explain the importance of producing pure enantiomeric compounds; explain why auxotrophic mutants and regulatory mutants are often used for production of amino acids by fermentation; describe common problems associated with amino acid fernentations and outline strategies that can be used to overcome them; describe the production of L-phenylalanine by direct fermentation from glucose, and explain how carbon balance analysis can be used to characterise the process; calculate the return on investment for a process, based on cost analysis data; compare and contrast the production of L-phenylalanine by precursor feeding and by bioconversion. Industrial production of amino acids by fermentation and chemo-enzymatic methods 273 A pendix 8.1 : Representation of and the nomenclature for s P ereo-isomers It is rather difficult to represent a tetrahedron which is threedimensional with a drawing of a formula that is two-dimensional. Two types of repxwntations are used, perspective and projection formulae. Fischer projections are widely used because of their simplicity. Y perspective projection Fischer formulae formulae projection You should note that on a Fischer projection the two formulae refer to the same molecule Nomenclature for stereo-isomers A system of nomenclature has been devised to describe optical isomers conveniently. These isomers differ in configuration and we have to be able to speafy the configuration at the asymmetric atom unambiguously. The original method had as a starting point the enantiomers of a standard compound, glyceraldehyde. One enantiomer rotated polarised light to the right or clockwise, and it was referred to as (+)-glyceraldehyde. The other rotated polarised light to the left or counterclockwise and was referred to as (-)-glyceraldehyde. The two enantiomers of glyceraldehyde were, according to their Fischer projections called D and L, from the Latin word dexter (right) and laevus (left). 274 Chapter 8 D lyceraldehyde L lyceraldehyde [( + -7 glyceraldehyde] [(hY ce raldeh yde)] The Dconfiguration is assigned to (+)-glyceraldehyde and the Gconfiguration is assigned to (-)-glyceraldehyde. Compounds related to D-glyceraldehyde were said to have a Dconfiguration while those related to L-glyceraldehyde were said to have a Lconfiguration (as is the case with naturally occming amino acids and carbohydrates). However, it is not always clear to which glyceraldehyde a given compound is 'related'. For example, although it is clear enough that the compounds are D rather than L, it is not clear whether is D or L. Because of the somewhat ambiguous definitions of D and L, another system of nomenclature was devised for asymmetric compounds, and it has largely replaced the old D, L-system since the 1960s. However, the D, L-nomenclature is still used for amino acids and sugars. The RS designation of chirality The absolute configuration of any chiral centre can be unambiguously specified using the RS notation. Industrial produdion of amino acids by fermentation and chemo-enzymatic methods 275 Figure A8.1 RS designation of chirality. 1) CHFClBr and 2) alanine. The stereoisomer of CHFClBr shown in Fip A8.1 has the R configuration, whereas that of alanine has the S configuration. The first step in using the RS designation is to assign a priority sequence to the four substituents by applying the rule that an atom with a higher atomic number has a higher priority than an atom with a lower atomic number. Hence, the priority sequence of these four substituents is a) Br, b) C1, c) F, and d) H, with Br having the highest priority. The next step is to orient the molecule so that the group of lowest priority d) points away from the viewer. For CHFClBr, this means that the molecule should be oriented so that H is away from us (behind the plane of the page). We then ask whether the path from a to b to c under these conditions is clockwise or counterclockwise. If it is clockwise (right-handed), the configuration is R (from the Latin rectus, right). If it is counterclockwise (left-handed), the configuration is S (from the Latin sinister, left). 276 Chapter 8 Now let us consider the RS designation of alanine (Table 8.2). The four atoms bonded to the a-carbon are N, C, C and H. The priority sequence of the methyl carbon and the carboxyl carbon is determined by going outward to the next set of atoms. It is useful to note the following priority sequence for biochemically important pups: -SH (highest), (lowest). Thus, the priority sequence of the four groups attached to the aarbon of alanine is a) -W+, b) -COO-, c) -CH3 and d) -H. The next step is to orient L-alanine so that its lowest priority group (-HI is behind the plane of the page. The path from a) -W+ to b) €00- to c) -CH3 is then counterclockwise (left-handed) and so L-alanine has an S configuration. -OR, -OH, -NHR, -N&, €00R, €00H, €HO, €&OH, C&j, B, -T, -D, -H Industrial production of amino acids by fermentation and chemo-enzymatic methods 277 Appendix 8.2 Examples of industrial production of amino acids by enzymatic methods A8.2.1 Enzymatic resolution of racemates Several classes of enzymes have been used to separate stereoisomers of a-H-and a-disubstituted amino acids, eg amidases, nitrilases, hydantoinases, acylases and esterases. A8.2.1.1 Amidases A very efficient and universal method has been developed for the production of optically pue L- and D-amino acids. The principle is based on the enantioselective hydrolysis of D,L-amino acid amides. The stable D,L-amino acid amides are efficiently prepared under mild reaction conditions starting from simple raw materials (Figure A8.2). Thus reaction of an aldehyde with hydrogen cyanide in ammonia (Strecker reaction) gives rise to the formation of the amino nitrile. The aminonitrile is converted in a high yield to the D,L-amino acid amide under alkaline conditions in the presence of a catalytic amount of acetone. The resolution step is accomplished with permeabilised whole cells of Pseudomonas putida ATCC 12633. A nearly 100% stereoselectivity in hydrolysing only the L-amino acid amide is combined with a very broad substrate specificity. 278 Chapter 8 Figure A8.2 The production of optically pure L- and D-amino add. (See text for further explanation). Industrial production of amino acids by fermentation and chemo-enzymatic methods 279 Not only the smallest optically active amino acid (alanine), but also leucine, several (substituted) aromatic amino acids, heterosubstituted amino acids (methionine, homomethionine and thienylglycine) and even an iminoacid, proline, are obtainable in both the L- and D-form. No enzymatic side effects are observed and substrate concentrations up to 20% by weight can be used without affecting the enzyme activity. The biocatalyst is used in soluble form in a batchwise process, thus poorly soluble amino acids can be resolved without technical difficulties. Reuse of the biocatalyst is in principle possible. A very simple and elegant alternative to the use of ionexchange columns or extraction to separate the mixture of D-amino acid amide and the Lamino acid has been elaborated. Addition of one equivalent of benzaldehyde (with respect to the D-amino acid amide) to the enzymic hydrolysate results in the formation of a Schiff base with the D-amino acid amide, which is insoluble in water and, therefore, can be easily separated. Acid hydrolysis (H2SO4, HX, HNO3, etc.) results in the formation of the D-amino acid (without racemization). Alternatively the D-amino acid amide can be hydrolysed by cell-preparations of Rhodococacs erythropdis. This biocatalyst lacks stereoselectivity. This option is very useful for amino acids which are highly soluble in the neutralised reaction mixture obtained after acid hydrolysis of the amide. Process economics dictate the recycling of the unwanted isomer. Path A in Figure A8.2 illustrates that racemisation of the D-N-benzylidene amino acid amide is facile and can be carried out under very mild reaction conditions. After removal of the benzaldehyde the D,Gamino acid amide can be recycled; 100% conversion to the Lamino acid is theoretically possible. Another method for racemisation and recycling of the Gamino acid (path B, Figure A8.2) comprises the conversion of the L-amino acid into the ester in the presence of concentrated acid, followed by addition of ammonia, resulting in the formation of the amide. Addition of benzaldehyde and racemisation by OH- (pH = 13) gives the D,L-amino acid amide. In this way 100% conversion to the D-amino acid is possible. A8.2.1.2 Nitrilases The bioconversion of a-aminonitriles, although up until now not used on an industrial scale, is of practical interest in the production of optical active a-amino acids. This, however, will only be the case if one can select a nitrilase that enantioselectively hydrolyses the aminonitrile. As illustrated in Figure A8.3 nitrilases catalyse conversions of nitriles directly into the corresponding carboxylic acids (route A), while other nitrile converting enzymes, the nitrile hydratases, catalyse the conversion of nitriles into amides (route B) which, by the action of amidases usually present in the whole cell preparations, are readily transformed into carboxylic acids (route C). 280 Chapter 0 Figure A8.3 Preparation of optically active a-amino acids via bioconversion of the corresponding a-aminonitriles. L-Amino acids could be produced from D,L-aminonitriles with 50% conversion using Pseudomonas putida and Brm'buctm'um sp respectively, the remainder being the corresponding D-amino acid amide. However, this does not prove the presence of a stereoselective nitrilase. It is more likely that the nitrile hydratase converts the D,L-nitrile into the D,L-amino acid amide, where upon a Lspecrfic amidase converts the amide further into 50% L-amino acid and 50% D-amino acid amide. In this respect the method has no real advantage over the process of using a stereospecific L-aminopeptidase (vide supra). A bacterial isolate APN has been shown to convert a-aminopropionitril enantioseledively to L-alanine (94% yield, Z% e e). However, the major disadvantage of this approach, is the low stability of most aminonitriles in water (for example a-aminophenylacetonitrile in water of pH 7, degrades completely within 48 hours). The aminonitriles are always in equilibrium with the aldehyde or ketone and ammonia/HCN. Polymerisation of hydrogen cyanide gives an equilibrium shift resulting in the loss of the aminonitrile. Therefore, a low yield in amino acids is to be expected, which makes this method less attractive for the industrial synthesis of optically active amino acids. A8.2.1.3 Acylases General Several L-amino acids are produced on a large scale by enzymatic resolution of N-acetyl-D,L-amino acids (Figure A8.4). Acylase immobilised on DEAE-Sephadex is for example employed in a continuous process while Degussa uses the free acylase retained in a membrane reactor. In the latter process the advantage of reuse of the enzyme and homogeneous catalysis are combined. However, the products are separated using ion-exchange columns and the starting material is a derivative rather than a precursor of the racemic amino acid, thus making the total process circuitous since it involves several chemical steps in addition to the enzymatic resolution step. Furthermore, racemisation of the unwanted isomer is not easily accomplished. Industrial production of amino acids by fermentation and chemo-enzymatic methods 281 Figure A8.4 Commercial process for the enzymatic production of L- and D-amino acids from N-acetyl-D,L-amino acids. Also the scope of the acylase method is limited in practice because chemical acylation of amino acids is difficult, and some N-acylamino acids are unsusceptible to the enzyme. Continuous process of amino acid resolution using immobilised acylase The Japanese firm Tanabe Inc Ltd has been operating, since 1969, the optical resolution hydrolysis of N-acyl-DL-amino acid by amino acylase which gives the L-amino acid and the unhydrolysed acyl-D-amino acid. To develop a continuous process, the immobilisation of aminoacylase of AspergdZus oryzae by a variety of methods was studied, for example ionic binding to DEAE-Sephadex, covalent binding to iodo-acetyl cellulose and entrapment in polyacrylamide gel. Ionic binding to DEAE-Sephadex was chosen because the method of preparation was easy, activity was high and stable, and regeneration was possible. The flow diagram of the enzyme reactor for continuous production of the L-amino acid is given in Figure A85 The acetyl amino acid is continuously charged into the enzyme column through a filter and a heat exchanger. The effluent is concentrated and the L-amino acid is crystallised. The acyl-D-amino acid contained in the mother liquor is racemised by heating in a racemisation tank, and reused. of DL-amino acids using aminoacylase. The principle is based on the asymmetn 'Cal 282 Chapter 8 Figure A8.5 Flow diagram for the continuous production of L-amino acids by immobilised aminoacylase. The advantages of the continuous process over the batch process are: savingofenzymes; 0 less human labour due to automation; increase in reaction yield, due to the easy isolation of L-amino acid from the reaction mixture. Due to these advantages the overall production costs for the immobilised continuous process were found to be 40% lower than that of the batch process. In Figure A8.6 a comparison is given between the batch process costs and the continuous production Costs. The plant is mainly used for L-methionine, L-valine, L-phenylalanine. The German firm Degussa AG uses immobilised acylase to produce a variety of L-amino acids, for example L-methionine (80, OOO tomes per annum). The principles of the process are the same as those of the Tanabe-process, described above. Degussa uses a new type of reactor, an enzyme membrane reactor, on a pilot plant scale to produce L-methionine, L-phenylalanine and L-valine in an amount of 200 tonnes per annum. Degussa *G Industrial production of amino acids by fermentation and chemo-enzymatic methods 283 Figure A8.6 Economic comparison of batch and continuous production of L-amino acids using amino acylases. a-Disubsituted amino acids are not easily prepared using acylases. Hog renal acylases I is unable to catalyse the hydrolysis of N-acetyl-a-methylphenylalanine. In contrast, carboxypeptidase A, is able to hydrolyse stereoselectively N-trifluoroacetyl-a-methylamino acids (Figure A8.7). Optically pure L-a-methylvaline and L-a-methylphenylalanine can be obtained using this method, but the use of an expensive trifluoroacetyl group makes it rather unattractive. Rgure A8.7 Preparation of optically pure a-methyl amino acids using carboxypeptidase A. A8.2.1.4 Hydantoinases Another option for resolution of optically active amino acids is illustrated in Figure A8.8. 284 Chapter 8 Figure A8.8 Enzymatic processes for the production of optically active a-amino adds via resolution of the racemic hydantoins. Chemically synthesised D,L-hydantoins prepared from the corresponding aldehydes via the Bucherer Berg reaction are converted by the bacterial cells (&lCillus bank), containing a D-specific hydantoinase, to a mixture of D-N-carbamoyl amino acid and L-hydantoin. The latter compound undergoes rapid and spontaneous racemisation under the conditions of the reaction, therefore, in principle 100% of the hydantoin is converted into the D-N-carbamoyl compound. The D-amino acid is obtained after treatment of the D-N-carbamoyl compound with nitrous acid. This process is operated on an industrial scale by the Japanese firm Kanegafuchi. An even more elegant approach for the production of D-phydroxyphenylglycine on an industrial scale uses the bacterium. Agrobacterium tvrdiobacfer (Figure A8.8). The organism is able to produce both D-hydantoinase and a second enzyme, N-carbamoyl-D-amino acid aminohydrolase, which catalyse the hydrolysis of N-carbamoyl-D-amino acid. The stereoselective hydrolysis of D,L-N-carbamoyl-methionine by the use of micro-organisms of the type Agobacterium rhiwgews IF0 13259, Corynebacterium sepedonicum IF0 3306, and Mycobacterium smegmatics ATCC 607, producing L-methionine and D-N-carbamoyl-methionine, has also been described. The stereoselective cyclisation of Ncarbamoyl-adisubstituted amino acids into the corresponding D-hydantoins is possible using bacteria belonging to the genera Aerobacter, Agrobacterium, Bacillus etc. D- and Ladisubsituted amino acids can be prepared from the D-hydantoins and the remaining L-Ncarbamoyl compounds, respectively (Figure A8.9). Although this mute seems quite elegant, there are some disadvantages. From the point of economic feasibility, it would be more attractive to start with a substrate that is a precursor of the optically pw adisubstituted amino acid. In the method described, first the D,L-amino acid has to be prepared and protected. After the enzymatic step depmtection is necessary to obtain the D-amino acid and the total process is cixuitous. Furthermore, the unwanted isomer cannot be racemised. Industrial production of amino acids by fermentation and chemo-enzymatic methods 285 Figure A8.9 Procedure for the preparation of optically adive u-disubstituted amino acids through stereoselective enzymatic cydisation of the Ncarbamoyl derivatives. A8.2.1.5 Esterases Alcalase selectively catalyses the hydrolysis of D,L-amino acid methyl and benzyl esters to pmvide L-amino acids and D-amino acid esters with high optical purity (Figure A8.10). Figure A8.10 Enzymatic resolution of D,L-amino acid esters. Microbial serine proteases, such as chymotrypsin, catalyse the hydrolysis of N-acetyl-L-amino acid esters (Figure A8.11). Figure A8.11 Enzymatic resolution of N-acetyl-D,L-amino acid esters. Numerous other examples of enantioselective hydrolysis of esters have been reported. For example chymotrypsin, immobilised in a liquid membrane of kerosene or cyclohexane, can be used for resolution of D,L-amino acid esters in an emulsion type 286 Chapter8 reactor. This emulsion-type enzyme membrane reactor is used principally for continuous racemate resolution. These are mapr disadvantage of the esterase resolution process. Since the optimum pH of the enzymic reaction is generally on the alkaline side, the esters used as substrates are non-enzymatically hydrolysed and the optical purity of the L-amino acids obtained is generally low. Also the substrate has to be protected at the amino pup in most cases in order to prevent formation of diketopiperasines. The esterase method is not attractive in practice and to the best of our knowledge is not used on an industrial scale. A8.2.2 Enzymatic asymmetric synthesis A8.2.2.1 Ammonialyases L-Phenylalanine can be synthesised from trm-cinnamic acid (Figure A8.12) catalysed by a L-phenylalanine ammda-lyase from Rhodococcus glutinis. The comm-tion of the process was limited by the low conversion (70561, low stability of the biocatalyst and the severe inhibition exerted by trans-cinnamic acid. These problems were largely overcome by researchers at Genex. The process, commercialised for a short period by Genex, involves a cell-free preparation of phenylalanineammda-lyase activity from Rhodotorulatllbm. Of industrial importance at present is the biotransformation of fumarate to Gaspartic acid by Escherichia coli aspartase. Modified versions have been developed, such as the continuous production of L-aspartic acid using duolite-ADSaspartase. A conversion higher than 99% during 3 months on a production scale has been achieved. Figure A8.12 Enzymatic conversion of franscinnamic acid to L-phenylalanine. L-alanine can be prepared from aspartic acid (Figure A8.13). L-Aspartate- decarboxylase produced by Xanthomonas qzue No 531 has been used to prepared L-alanine in 95% yield from 15% L-aspartic acid solution. Other strains, ie Pseudomonas dacunhae or Achromobacter patifer, give comparable yields of L-alanine. The process has been conu-nercialised by Tanabe. Industrial produdion of amino acids by fermentation and chemo-enzymatic methods 287 Figure A8.13 Two step enzymatic synthesis of L-alanine from fumaric acM. A8.2.2.2 Industrially operated enzymatic synthesis Enzymatic synthesis of Gaspartic acid and Galanine L-aspartic acid has been produced on an industrial scale by the Tanabe Seiyaku Co Ltd, Japan, in a batchwise process using whole cells of Escherichia cdi with high aspartase activity. In this process, L-aspartic acid is produced from fumaric acid and ammonia using aspartase, as described in Figure A8.13. To develop a continuous process, the immobilisation of the as artase was studied, but production. To overcome these disadvantages the E. coZi cells were directly immobilised by entrapping them in polyacrylamide gel. For the industrial application, a continuous aspartase reactor system, using a column packed with the immobilised E. mZi cells, was designed. The column used for the industrial production was designed as a multi-stage system with heat exchanger, because the reaction is exothermic. The half life of the conventional polyacrylamide gel immobilised E. coli was 120 days at 3yC. E. cdi cells immobilised with Xcarrageenan showed a half life time of 680 days at 3yC. From 1978 the carrageenan method has been used to immobilise the E. coli. L-alanine can be produced from L-aspartic acid using L-aspartate4decarboxylase. This reaction has been described in (Figure A8.13). L-alanine has been produced since 1965 by a batchwise enzyme reaction, using the L-aspartase- decarboxylase of Pseudaumas dacunhae. Again a continuous process was developed using Pseudomonas dacunhae immobilised with x-carrageenan. One of the problems in the continuous production is the evolution of carbon dioxide during the reaction. This carbon dioxide gas makes it difficult to obtain plug flow conditions in the reactor. Therefore, a closed column was designed, which allows the enzyme reaction to perform at high pressure (about 4 bar). The flow diagram of the closed column reactor is shown in Figure A8.14. The carbon dioxide remains dissolved in the reaction mixture, obtaining almost plug flow conditions. The efficiency of the immobilised cells under high pmsure is 42% higher than the conventional column system. the operational stability of the preparations was not su Ki cient for industrially 288 Chapter 8 Figure A8.14 Closed column reactor for the production of L-alanine. 1) reador; 2) plunger pump; 3) pressure control valve; 4) reservoir; 5) substrate tank; 6) pressure gauge; 7) safety valve; 8) heat exchanger. Aspartic acid may be produced continuously from ammonium fumarate using immobilised E. coli. L-alanine is continuously produced using immobilised Ps. dacunhae. If both E. coli and P. dacunhae are employed simultaneously, L-alanine could be produced more efficiently from ammoNum fumarate according to the reaction shown in Figure A8.13. To prevent the formation of byproducts like L-malic acid and D-alanine, the cells undergo a pH-treatment tu inactive fumarase and alanine racemase. Several reactor conformations have been investigated, but a two reactor system was found to be the most effective. The flow sheet of this two reactor system is given in Figure A8.15. In the first reactor Gaspartic acid is formed, which reacts in the second reactor to L-alanine. Industrial production of amino acids by fermentation and chemo-enzymatic methods 209 Figure A8.15 Flow diagram for the continuous production of L-aspartic acid and L-alanine. A8.2.3 Oxidation - Reduction reactions A8.2.3.1 Aminotransferases A new development is the industrial production of L-phenylalanine by converting phenylpyruvic acid with pyridoxalphosphatedependent phenylalanine transaminase (see Figure A8.16). The biotransformation step is complicated by an unfavourable equilibrium and the need for an aminodonor (aspartic acid). For a complete conversion of phenylpyruvic acid, oxaloacetic acid (deamination product of aspartic acid) is decarboxylated enzymatically or chemically to pyruvic acid. The use of immobilised E. coli (covalent attachment and entrapment of whole cells with polyazetidine) is preferred in this process (Figure A8.17). 290 Chapter 8 Figure A8.16 Production of L-phenylalanine via phenylalanine transaminase. Figure A8.17 ldealised structure of a representative polyazetidine used for the irnrnobilisation of whole cells. Polyazetidine prepolymer may be cross-linked in aqueous solution by reaction with amine, thiol, hydroxyl, carboxylic acid or other polyazetidine groups. Cross-linking occurs upon water removal, heating or by changing to a basic pH. The immobilised cell/polymer composition may be prepared in the form of membranes, fibres, tubes or beads. Industrial production of amino acids by fermentation and chemo-enzymatic methods 291 The industrial development of biotransformations is hampered currently by lack of commercial availability of biocatalysts at a reasonable price, insufficient operational stability of most biocatalysts and the practical problems associated with the exploitation of cofactordependent biocatalysts. Many procedures have been suggested to achieve efficient cofactor recycling, including enzymatic and non.enzymatic methods. However, the practical problems associated with the commeraal application of coenzyme dependent biocatalysts have not yet been generally solved. Figure A8.18 illustrates the continuous production of L-amino acids in a multi-enzyme-membrane-reactor, where the enzymes together with NAD' covalently bound to water soluble polyethylene glycol 20,000 (PEG-20,WNAD) are retained by means of an ultrafiltration membrane. Figure A8.18 A racemic mixture of u-hydroxyacids (like L, D-lactate) can be transformed via the corresponding u-ketoacid (pyruvate) to the desired L-amino acid (L-alanine) with cofactor recydi ng. So far the economic feasibility of co-enzyme dependent biocatalyses is confined to datively small market niches comprising products with high added value.