ABSTRACT
It is more than 40 years ago that studies on the metabolism of microorganisms revealed the existence of isoenzymes in the allosteric regulation of branched pathways. Their role in the synthesis of amino acids derived from aspartate was especially well investigated in Escherichia coli, particularly from the point of view of their allosteric properties (Patte et al. 1964). The enzyme aspartokinase catalyses the phosphorylation of the amino acid aspartate, that being the fi rst step in the biosynthesis of three different aminoacids, threonine, methionine and lysine, as well as of a fourth, isoleceucine (from threonine). Animals lack these pathways, which is what makes these amino acids ‘essential’ for us – we have to get them from our diet. At that time the global regulation of entire metabolic networks had not been clarifi ed in a quantitative sense. Everyone seemed to take it for granted that there must be a logical reason behind the existence in E.coli for instance of three aspartokinases, each of them being regulated, meaning inhibited and/or repressed (at the level of gene activity), by its ‘corresponding’ amino acid. This was subsequently also found in plants but with a more complex pattern of regulation. It was not clear whether the more complex regulatory pattern refl ected the necessity of a multiplicity of metabolic responses or simply a ‘stratifi cation’ of different regulations during evolution. In other words, could a simpler system of regulation offer the same wealth of metabolic responses? The pattern in the thale cress Arabidopsis thaliana has been painstakingly deciphered over a ten-year period by Curien and co-workers; all the steps, including their regulatory properties, are known in detail. As shown in their most recent publication (Curien et al. 2009), today such questions can be answered by building a mathematical model to simulate the global functioning of the network. In order to obtain a steady state in the model as it occurs in vivo, it is necessary to add to the network the cellular demands of lysine, threonine and isoleucine in protein synthesis. Methionine is not a variable of the system because it is the precursor of S-adenosylmethionine (AdoMet) which in addition plays a regulatory role in the network (see below). The AdoMet concentration is set at its physiological value. The amino acid demand is adjusted in order to obtain a concentration of intermediate metabolites close to the ones measured in vivo. This steady state is called the reference state and is the starting point of all the other simulations. The authors use metabolic control theory (MCT, also known as Metabolic Control Analysis) (Kacser and Burns 1973; Heinrich and Rapoport 1974; Reder 1988) to identify the more sensitive steps and metabolites. In metabolic networks, controls at the ‘supply’ and ‘demand’ ends usually act in opposition (Hofmeyr and Cornish-Bowden 2000); most trivially, an increase in substrate tends to increase the fl ux through a pathway whereas an increase in the product will tend to decrease the fl ux. In the present case, most of the control is on the ‘demand’ steps in the amino acids’ biosynthetic pathways: that allows the network to respond directly to a change in protein synthesis. There is, however, a high control coeffi cient associated with the aspartate kinase isoform AK1, which is at the ‘supply’ end. The consequence is a signifi cant contribution of AK1 to the maintenance of a threonine steady-state. This, according to the authors, is one of the reasons why the threonine level is less stable than those of the other amino acids – both in the model and in planta. The fi rst important result obtained from the model is the calculation of the fl ux through the different branches of the network and their detailed assignment to the different isoforms. It appears from fi gure 3 of the paper (which very clearly shows the different fl uxes) that with regard to the fi rst (aspartate kinase) step catalysed by the four aspartokinases isoforms in A. thaliana, the fl ux is mainly accounted for by one of them, namely AK1 (73%). The second important result is that to a large extent the different pathways behave independently. This means that when, for instance, the demand for threonine is increased, lysine production is not affected even though the two biosynthetic pathways share common steps in the beginning. This nonintuitive result is due to the special features of the regulation pattern, particularly due to the presence of the.