- Email: [email protected]
Flux analysis and control of the central metabolic pathways in Escherichia coli
ELSEVIER
MICROBIOLOGY REVIEWS FEMS Microbiology Reviews 19 (1996) 85-116
Flux analysis and control of the central metabolic pathways in Escherichia coli Harry Holms ~ Robertson Institute of Biotechnology, 54 Dumbarton Road, Glasgow GI 1 6AQ, UK
Received 15 April 1996; revised 18 July 1996; accepted 19 July 1996
Abstract
The growth of the bacterial cell involves the co-ordination of the fluxes of carbon into a considerable diversity of products that are the components of the cell. Fortunately the monomers from which the cell's polymers are made are themselves synthesised from a relatively small group of precursors that are the products of the central metabolic pathways. This simplification renders cell metabolism accessible to flux analysis, a method for handling experimental data to derive metabolic fluxes. Through such analysis of the growth of Escherichia coli ML308 on 11 single carbon sources in batch, turbidostat or chemostat culture general patterns are discernible. Most significant among these are that growth on different carbon sources is achieved without any obvious enzyme acting as a regulator of metabolic flux, except when acetate is the sole source of carbon. In this case a junction is created at which isocitrate dehydrogenase (ICDH) and isocitrate lyase (ICL) compete for their common substrate and this competition is resolved by partial inactivation of ICDH to match flux through ICL and this balance limits growth rate. In this sense, flux through ICDH and ICL is 'rate-limiting'. Uptake of six of the remaining carbon inputs exceeds the capacity of the central metabolic pathways (CMPs) to sustain flux to the precursors required for growth and the CMPs are balanced by excretion of acetate. Restriction of carbon uptake by chemostat progressively diminishes growth rate and acetate excretion until acetate excretion is prevented. For the four remaining carbon sources, uptake is apparently restricted and the products are biomass, carbon dioxide and water. Carbon sources feeding the phosphorylated parts of the CMPs flux relatively more carbon to precursors (Pre-C) than CO 2 when compared with carbon sources which teed into the non-phosphorylated pathways. P r e - C / C O 2 ratios for the former are 1.73-3.91 and for the latter are 0.46-0.78. Flux analysis of all 11 carbon sources shows that there is an overabundant supply of 'energy' (ATP + [2H]), generated by the CMPs, in all phenotypes and conditions down to a glucose chemostat at /x of 0.72. This excess energy is a thermodynamic inefficiency which must be dissipated as heat. E. coli ML308 probably evolved in circumstances of 'feast" and 'famine'. The two strategies selected (excretion of surplus carbon and restriction of /~) would appear to be defences against 'feast'. Presumably there are defences against 'famine'. These are not made obvious by flux
Abbreviations: Ac.CoA, Acetyl CoA; C3P, Triose phosphate; C4P, Tetrose phosphate; C5P, Pentose phosphate; EDP, Entner-Doudoroff pathway; ETS, Electron transport system; G6P, Glucose 6-phosphate; ICL, isocitrate lyase; ICDH, isocitrate dehydrogenase; ME, Malate enzyme; OAA, Oxalacetate; OGA, Oxoglutarate; PDH, Pyruvate dehydrogenase; PEP, Phosphoenolpyruvate; PEPC, Phosphoenolpyruvate carboxylase; PEPsyn, PEP synthetase; PK, Pyruvate kinase; PPP, Pentose phosphate pathway; PTS, Phosphotransferase system; MDH (de), Malate dehydrogenase (decarboxylating); PEPCK, Phosphoenolpyruvate carboxykinase; PG, Phosphoglycerate; Pyr., Pyruvate * Corresponding author. Tel: +44 (141) 330 5264; Fax: +44 (1411 330 4629. (1168-6445/96/$32.00 Copyright © 1996 Federation of European Microbiological Societies. Published by Elsevier Science B.V. PII S01 68-6445(96)(/(1(126-5
86
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
a n a l y s i s but allosteric control o f irreversible e n z y m e s w o u l d protect pools o f essential nutrients f r o m rapid depletion on the sudden onset of 'famine'.
Keywords: Flux analysis; Central metabolic pathway; Control; Escherichia coli
Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
87
2. M e t h o d o l o g y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Definitions and units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. F r o m inputs to precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Flux analysis o f g r o w t h on g l u c o s e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Flux analysis o f g r o w t h on glycerol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Flux analysis o f g r o w t h on p y r u v a t e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Flux analysis o f g r o w t h on acetate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7. Flux analysis o f g r o w t h in c h e m o s t a t s and on other carbon s o u r c e s . . . . . . . . . . . . . . . . . . . . . . . . . . .
88 88 89 9I 94 95 95 97
3. D i s c u s s i o n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. T h e distribution o f carbon fluxes in the C M P s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Intervention to reduce acetate excretion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. E x c e s s i v e uptake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. E x c e s s i v e flux t h r o u g h P D H . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. C o n v e r s i o n o f A c . C o A to acetate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.4. E n h a n c e m e n t o f flux to O A A and O G A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Controls and limitations in the C M P s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1. E x c e s s i v e flux o f inputs c a u s e s acetate excretion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2. B a l a n c e b e t w e e n input a n d output f r o m the C M P s d e p e n d s on control o f input . . . . . . . . . . . . . . . . 3.3.3. R e g u l a t i o n o f g r o w t h on acetate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4. F l u x e s to e n e r g y c o n s e r v a t i o n in the C M P s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.5. M a i n t e n a n c e e n e r g y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.6. T u r n o v e r o f pools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.7. Control o f the C M P s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.8. Control theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.9. D e d u c t i o n s f r o m flux analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
100 100 102 102 103 103 104 104 105 105 106 108 110 1 I0 I 12 113 114
Acknowledgement .............................................................
115
References
115
.................................................................
87
H. Holms / FEMS Microbiology Reviews 19 (1996) 85- I 16
1. Introduction
This review describes the fluxes in Escherichia coli growing on a variety of single carbon sources in batch, turbidostat or chemostat cultures. Virtually all the data refer to one particular strain (ML308) from the collection of the late Jaques Monod which has been deposited as ATCC 15224. It is constitutive for /3-galactosidase by virtue of a mutation in the repressor gene but otherwise presumed to be wild-type. Flux analysis is a description of the metabolic events which convert feedstocks to biomass and by-products and is derived from a precise measurement of utilisation of inputs, generation of biomass and excretion of by-products. This review will only consider carbon fluxes but fluxes of nitrogen, sulfur, phosphate, etc. can be defined in the same way. In principle, the conversion of a carbon source to biomass is simple (Fig. l). Each individual carbon source is taken up by the cell and fed into the CMPs. The organisation of CMPs is covered in many standard texts, of which some [1-3] are more userfriendly than others. The CMPs contain some 30 or so compounds all of which are phosphorylated or are carboxylic acids, several are both and two are thiol esters of Coenzyme A. The interconversions of intermediates are catalysed by enzymes organised in pathways which are glycolysis (Emden-Meyerhof pathway: EMP), the pentose phosphate pathway (PPP), pyruvate dehydrogenase (PDH), phosphoenolpyruvate carboxylase (PEPC) and the Krebs cycle or variations on these (e.g. Entner-Doudoroff
Carbon Source
pathway (EDP), phosphoketolase pathway) depending on the chemical nature of the carbon input Which dictates the point of entry into the CMPs. The number of enzymes involved is larger than the number of substrate to product conversions because some reactions have more than one enzyme available and many reactions use different enzymes depending on which direction they operate (but many reactions are freely reversible). Many reactions involve cofactors (NAD(P)+/NAD(P)H, H +, F A D / F A D H , A T P / A D P / A M P , C o A / A c . CoA and inorganic ions such as Mn 2+ and phosphate) and, in any steady state, these are recycled and maintained in constant pools. It is thus possible to consider fluxes of carbon in isolation. Glucose is taken as the standard to which other inputs are compared and feeds CMPs consisting of EMP, PPP and the Krebs cycle. A fraction of the intermediates of the CMPs are the precursors for biosynthesis and each precursor is the starting point for synthesis of monomers which, in turn, are polymerised and then organised into the polymeric structures of the biomass. These anabolic processes require energy input in the form of ATP and reducing power [2H]. These pools of 'high-energy' compounds are maintained by the CMPs which thus perform a dual (amphiboli~) function. The catabolic components of the CMPs are oxidative decarboxylation of pyruvate (Pyr), isocitrate, OGA and 6-phosphogluconate to generate carbon dioxide. The oxygen input is water with the result that there is a much greater generation of [2H] than carbon dioxide. Molecular oxygen is not utilised in the CMPs although they are able to oxidise carbon sources completely. Thus for glucose, total oxidation follows the equation: C6HI206 +
H20 Heat
)~
entr , Metabolic Pathways
I~ATP
tile cycle~PiOH R Precursors PiOH~~ X'OP~ADP~ ~eat !m Acetate H20 ~X=OJ ~ ATP.~[ Mon/ ers 2 H~O Polymers ~v Siomass
Fig. I. Functions of the central metabolic pathways (CMPs).
6H~O-~
6C02 +
1212H]
The reducing power thus generated is then oxidised in the electron transport system for which the terminal electron acceptor is molecular oxygen. The pools of reduced cofactors are kept constant by flux of electrons (and associated protons) through the electron transport system (ETS) in which the terminal electron acceptor is molecular oxygen. The flow of electrons down the thermodynamic gradient of the ETS is coupled, by the proton motive force, to the phosphorylation of ADP to ATP.
88
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
The uptake of glucose into the CMPs must satisfy all these fluxes derived from the CMPs. The fundamental requirement asked of the CMPs is the provision of precursors in exactly the amounts required to sustain fluxes to monomers which, in turn, is specified by the DNA and the flux to polymers defined by the expression of that fraction of the DNA specific to the glucose phenotype. In other words, the relative fluxes from precursors are predetermined and their conversion to biomass necessarily also determines the demands for ATP and [2H] for their anabolism. As a minimum, the uptake of carbon source must satisfy these demands and, if it does this, it either matches them precisely or limits their rate, i.e. limits growth rate (tz). However, in the particular case of glucose, uptake delivers carbon into the CMPs at a rate which exceeds the requirements of flux to precursors and energy. The total pool of solutes in E. coli is constant [4] and the cell wall is designed to contain the consequent osmotic pressure but it could not sustain the much higher pressures which would result from accumulation of excess carbon as low molecular weight compounds. It follows that excessive uptake of carbon source must be accommodated by other mechanisms of which there are three: intermediates in excess of requirements can be converted to storage polymers [5,6]; the coupling of ETS to phosphorylation may operate at less than maximal thermodynamic efficiency [7-10]; and the surplus carbon fluxing through the CMPs can be excreted as low molecular weight compounds [4,11-13]. E. coli ML308 does not accumulate polymers when flux to amino donors is sufficient and, as is shown below, some excess carbon is oxidised to generate heat and the remainder is excreted, usually as acetate. How can the fluxes of this system, and others, be measured and described?
2. Methodology The approach is simple. When you measure utilisation of carbon source, production of biomass and other products and growth rate then, provided you also know the metabolic routes which operate, you
have all the data required to construct a quantitative description of all the reactions (net) which sustain these processes. Such a description contains the flux through every enzyme in the CMPs and is called a flux analysis. 2.1. Definitions and units The basic parameter of a fermentation is the biomass, expressed as dry weight-usually kilogram dry weight biomass (kg dry wt). In the end, everything else is related to this. All other measurements are made in moles which avoids the confusion of changes in weight when substrates differ from product by virtue of phosphorylation, oxidation, decarboxylation, etc. An 'input' is the amount of a feedstock which is taken up and metabolised by the biomass. Thus, when E. coli is growing on glucose as the sole carbon source, we measure the input as the amount of glucose required to generate 1 kg dry wt. Throughputs are also amounts (in the same units) which are defined as the amount of a substrate (e.g. fructose 1,6-bisphosphate) converted to products (the triose phosphates: C3P) in generating 1 kg dry wt. Precursors are converted to monomers contained in the end-product (biomass) and both are in the same unit. There are, of course, other products such as CO 2 and sometimes other excreted products which also are expressed in the same units. In the steady state, the biomass generates itself at a constant rate described by the growth rate /x (per hour). The product of the amounts defined above and /z are m o l / k g dry w t / h and these are fluxes which describe the rate at which an input is used, the flux by which a substrate is converted to product or a monomeric constituent of biomass is generated or another product such as acetate is excreted. The last unit of possible significance in flux analysis is pool size which is an amount expressed as moles of an intermediate contained as a pool within the biomass and again has the unit m o l / k g dry wt in any given steady state. The rate at which the intermediate must be generated to maintain pool size is derived from /x in the same way to give moles pool intermediate/kg dry w t / h . In prokaryotes where pool sizes are very variable and usually small relative to their throughput, this value has limited importance and the turnover of the pool is probably more significant.
89
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-1 / 6
This is the rate of input or output (which are the same) divided by the pool size, i.e. m o l / k g , dry w t / h / m o l / k g • dry wt which has the dimension of a number/h and is the number of times the pool turns over every hour. It is obvious that in a linear segment of metabolism where the flux through each enzyme in the chain is the same, then the turnover of each pool of intermediates is inversely proportional to the size of the pool (this is discussed in an example below, Section 3.3.6.).
Glucose6P ,
M°D~ii! cetone:r u~'6~ilcera~ldehyde3P
/
DiPGlycerate
2.2. From inputs to precursors
3PGlycerate~ >Monomers
E. coli ML308 converts a single carbon source to biomass, carbon dioxide and, sometimes, acetate.
Glucose
h~Monomers
~ Glucose6Pi ~Monomers
2PGlylerate Monomers *~ Monorners~
~PenolPyruvate
Py ~PC /Malate ~4 ~
,,JPyruvate
Oxalac( :ate
Fumarate
h2~lonomers
F;iiioiil;; z i' DihyxyacetoneP -~-~lyceraldehyde3P Monomers
DiPGlycerate 3PGlycerate~ ~,Monomers 2PGlylerate
Monomers< Monomers <~
Pyruvate
~Penol
,Pyruvate
OxalaeE tate
Fumarate
M....... <~D~coA 'J~C i~1ate~M°..... Succ!nate iso
itrate Sy!ICoA '~ Glutarate, Oxo ~ Monomers
Fig. 2. Pathwaysto precursorsof monomersduringgrowthon
glucose. PDH: pyruvate dehydrogenase;PEPC: phosphoenol pyruvatecarboxylase;PK: pyruvatekinase;PPP: pentosephosphatepathway;PTS:phosphotransferasesystem.
M....... <~D~coA ~Ci] 'ate ~M . . . . . . . Succmate ; iso
Citrate Su~ylCoA ~P~Glutarate Oxo [ =~Monomers
Fig. 3. Pathwaysto precursorsof monomersduringgrowthon glycerol. The CMPs receive the carbon input and direct the flux to all the outputs. The large number of monomers from which biomass is assembled are made by known biosynthetic routes served from a rather limited number of precursors. These are usually the same, irrespective of the carbon source because the biosynthetic route to any given monomer usually starts from the same precursor. However, there is one exception in that the routes to C 4 and C 5 sugars are taken from the PPP which can be fed either by oxidative decarboxylation of 6-phosphogluconate or by rearrangement of C3P. In addition, the fluxes through the CMPs are modified to accommodate provision of precursors from different inputs, for example, gluconeogenesis from pyruvate requires a unique enzyme to generate phosphoenol pyruvate (PEP) and then reversal of glycolysis (using fructose
90
H. Holms
/ FEMS Microbiology Reviews 19 (1996) 85-116
bisphosphatase rather than phosphofructokinase) to generate glucose 6-phosphate (G6P). However, as far as carbon fluxes are concerned, a principal function of the CMPs is to sustain fluxes to the precursors for biosynthesis and these must be considered relative to the carbon input. The first task for flux analysis is to construct a diagram of the CMPs for each carbon input under study. Such diagrams are shown for glucose (Fig. 2), glycerol (Fig. 3), pyruvate (Fig. 4) and acetate (Fig. 5). The pathways from precursors to monomers are found in the standard texts and, more particularly, in an earlier systematic treatment of the basic principles of flux analysis [4]. The precursors used [4] to make biomass are G6P, pentose phosphates (C5P), tetrose phosphate (C4P), glyceraldehyde, 3-phosphate (C3P), phosphoglycerate (PG), PEP, Pyr, Ac.CoA, oxaloacetate (OAA)
Glucose6P~
~
s
' ~,M. . . . . . .
Monomers
/
,
Monomers
PenolPyruvate~ S~nE~if~ E P ( /Malate ~
x
Pyruvate
//PO.
Oxalac ate
Monon,ers*
AeetylCoA
.~Monomers
2PGly!erate
Monomers
<~
PenolPyruvate
Pyruvate
\
/MO.(dc)
/ Ae?tyIP
/
Citrate :1" Glyoxylate A ~ ICLY
isoC~ra te~
_Succunate
Succi!ylCoA
~P~GluOtX°ate/,
M. . . . . . .
.~Monomers
2PGly(.=rate
~P~
/
DiPGlycerate
Fig. 5. Pathways to precursors of monomersduring growth on acetate. ICL: isocitrate lyase; MDH (dc): malate dehydrogenase (decarboxylating); ME: malate enzyme; PEPCK: phosphoenolpyruvatecarboxykinase.
DiPGI' ,eerate
External Pyruvate
~ -D~Glyeeraldehyde3P
3PGlycerate [
Acetate
rU ~ G I I I I c ~ hyde3P
3PGly :erate
roxyacetoneP
. ....... :'=%cet,
F t t 16d,P ~
Monomers
~Monomers
Fruct( ~l.6diP/'
Fruct!se6P%~hf~
DihydroxyacetoneP
se6P%~l
Fruc'
~ ~Monomers
Glucose6P
~ Monomers
Fumarate
M...... s
Suc( ~ate
Cit 'ate
Monomers iso Citrate
Succi
ICoA
\oxo/r ~ Glutarate
Monomers
Fig. 4. Pathways to precursors of monomersduring growth on pyruvate. PEPsyn:phosphoenolpyruvatesynthetase.
and OGA. All of these precursors must always be provided to generate new biomass and our four models achieve this by minor modification of the CMPs and unique mechanisms to take up the particular carbon source and deliver it into the CMPs. These can be briefly described (for pathways see [1-4]). Glucose is taken up and phosphorylated by the phosphotransferase system (PTS). Glycerol (Fig. 3) is taken up by facilitated diffusion, gives G6P by gluconeogenesis but can feed the PPP from C3P without oxidative decarboxylation. Pyruvate (Fig. 4) is taken up by a proton symport and requires PEP synthase to generate PEP. Acetate (Fig. 5) is taken up by a proton symport, phosphorylated by acetokinase to acetyl phosphate and thence to Ac.CoA by a phosphotransacetylase. All other precursors, in the acetate
H. Holms / FEMS Microbiology Ret,iews 19 (1996) 85-116
pbenotype, depend on the glyoxylate bypass which generates malate and thence OAA from Ac.CoA by ICL and malate synthase (Fig. 5). How do we measure the fluxes from precursors to biomass? OGA is the precursor for four amino acids: glutamate, glutamine, proline and arginine. The content [4] of these amino acids in glucose-grown E. coli ML308 is (tool kg - ~ dry wt) Glu 0.353, Gln 0.201, Pro 0.252 and Arg 0.252 and as each derives from 1 tool OGA it follows that an output of 1.058 tool OGA is required to make 1 kg dry wt biomass. At a /z of 1.0 the flux of OGA to biosynthesis is also 1.058 mol kg-~ h -~. In the same way, the output of all the other precursors is calculated from the monomeric composition of the biomass and the growth rate (Table 1) on any different carbon sources from measured data [4,13,14].
91
" Gluelose6P~.'.'.'.'.'.'.'.'.~M.......
TS•
Glucose
Fructose6P
Fru~bisP TriosePs~
Monomers
¢-7
~
Monomers
yruvate~
Monomers
3PGlycerate
Peno
Monomers~
Pyruvate
Oxal ~
Flux analysis is simply a method of organising data and these are taken, in the first instance, from Table 1 and applied to the metabolic maps defined in Fig. 2, Fig. 3, Fig. 4 and Fig. 5 for flux of carbon only. All cofactors are taken from pools which, in the steady state, are maintained constant by regeneration and. as there is no net change, are ignored. Furthermore, as the flux through each step of a linear segment is the same, many of these are omitted and only where fluxes are changed by removal of precursors or excreted products or at branch points do they need to be included. The conversion of data to a flux
Monomers
Acetate
2.3, Flux analysis of growth on glucose
M....... ~ r ~ . ~ _ A~:~
__L__ C.!ate ,
,
AOP~ArPAcetaisoteCitrao~o te
o,o,.rt&
Excrete
Monomers
Fig, 6. Flux diagram for growth of E. coli ML308 on glucose showing only those fluxes directly derived fiom measured data (mol/kg dry wt/h).
Table I Fluxes (mol/kg dry w t / h ) from precursors to monomers from six carbon inputs at various growth rates (/x) Precursor Glucose 6(P) Triose(P) (P)Glycerate (P)enol pyruvate Pyruvate Oxalacetate Oxoglutarate Acetyl CoA
Carbon source (#)
(I,00)
Glucose (0.94)
Gluconate (0.90)
Fructose (0.72)
Glycerol (0,70)
Fumarate (0.63)
Acetate (0.43)
1.026 (1.984) * 0.140(1.636) + 1,362 0.554 2.318 1.688 1.058 2.643
1.86 ~ 0.13 1.28 0.52 2.18 1.59 0.99 2.48
0.92 ~ 0.13 1.23 0.50 2.09 1.52 0.95 2.38
0.74 1.18 + I).98 0,40 1.67 1.22 0.76 1.90
0.72 1.15 + 0.95 0.39 1.62 1.18 0.74 1.85
0.65 1.03 + 0.86 0.35 1.46 1.06 0.67 1.67
0.44 0.70 (}.59 0.24 1.00 0.73 0.45 1.14
Flux to PPP from G6P. + Flux to PPPs from triose(P) (see Fig. 18),
92
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
TS•
diagram requires three stages. Firstly, a diagram specific to the carbon input and the routes to precursors must be constructed. Secondly, the measured data are expressed as fluxes for input and provision of precursors for monomer biosynthesis and entered into this diagram. The simplest basic flux diagram for growth on glucose incorporates these two stages and includes input, output of precursors, acetate excretion and the inevitable consequences of these measured data-the flux of PEP to pyruvate for glucose uptake by the PTS and the anaplerotic flux to the Krebs cycle for provision of OAA and OGA (Fig. 6). This figure illustrates the principles of flux analysis which can be summarised as follows: when you measure everything that goes into the CMPs and everything that comes out, then you can construct a flux diagram provided you know the metabolic routes used. The fluxes are written in boxes as mol/kg dry wt./h. The flux diagram is then completed (Fig. 7) by simple arithmetic which is very straight forward and will be illustrated by only a few salient points. The flux of glucose into the cell is 10.57 of which 1.86 fluxes to monomers leaving 8.71 for glycolysis. Aldolase splits FDP to 2 mol of C3P of which 0.13 is used as precursor for lipid biosynthesis. A further 1.28 is removed to monomers as PG leaving 16.01 as the flux to PEP. Of this amount 10.57 is fluxed to pyruvate to donate phosphate for glucose uptake, 0.52 to biosynthesis of aromatic amino acids, 2.58 for anaplerotic provision to the Krebs cycle leaving
= Glucose6P "~'-~l~-~Monomers I
Glucose
Fructose6P
Fru~bisP TriosePs ~
Monomers
3PGlycerate ~
Monomers
PenolF~yru~at~e M. . . . . . . ~ l ' ~ ' 8 - ~ P y r u y ate
M. . . . . . . " ~ - ~ ' - -
~""'~"
Monomers
AcetateOXal~
t t r~j
M. . . . . . .
~
AOPATPAcet ~ ateisoCitrao~o te Glutarate Excrete
Fig. 7. Complete flux analysis of growth of E. coli ML308 on glucose. Fluxes are mol/kg dry wt/h.
2Oxoglutarate ,coo.
Glutamate coo.
C=Oi
CHNH2
COOH NA
+
COOH AcCoA
Glutamine
,coo.
coo.
N-AcetylgJutamyl semlaldehyde c,~) coo.
CHNH-COCH 3
CHNH-COCH 3
j/~ ~ CHNH_COCH 3
N*Acetylglutamate N-Acetylglutamyl(P)
CoA
COOH ATP
ADP COO(P) NADPH,H + NADP+ CHO
Proline OGA
COOH
F. . . . . .
COOH COOH
~..,
.~H,
CH2NH-- ICI--NH2 NH Arglnlne
CH2NH--C:NH
~ A s p
.NH
Argininosuccinate
COOH (~) ~.,
AMP (P'P)
ATP CH2NH.CONH 2 Citrulllne
COO/P ) COOH Acetate COOH CHNH2 L CHNH-COCH, ~,H2 < ¢,H, CH2NH2 Ornithine
Fig. 8. Synthesis of the glutamate family of amino acids.
CH2NH2 N-Acetylornithine
H. Holms / FEMS Microbiology ReL'iews 19 (1996) 85-116
Carbonsource OXOgi#/u~'u,rPar,:c ur. . . . C
~
D
P
.
Arginine
6ATP AMP 4NADPH,H+ "Energy" 5(P) NAD+ (P.P) 4NADPH,H ÷ ~ A D H H ,
Carbon source ÷ 02
÷
~ Carbon dioxide + H20
Fig. 9. Net reactions in CMPs during synthesis of arginine from oxoglutarate.
2.34 to be converted to pyruvate by pyru~zate kinase (PK). The flux to pyruvate is 12.91 (10.57 + 2.34) of which 2.18 is used for biosynthesis. The remainder (10.73) is decarboxylated by PDH to Ac.CoA of which 2.48 is used for biosynthesis, 4.89 fluxed to acetate excretion and the remainder (3.36) fluxed into the Krebs cycle by citrate synthase which also uses 3.36 m o l / k g dry w t / h of OAA made up of the fluxes to OAA (2.58 by PEPC and 2.37 by operation of the cycle) less 1.59 removed for biosynthesis (2.58 + 2.37 - 1.59 = 3.36). This diagram (Fig. 7) describes all the fluxes through the CMPs when E. co/i ML308 is growing on glucose. Because we know the biosynthetic routes from precursors to monomers it is possible to extend the flux analysis to each monomer. This would generate a very large flux diagram and only one will be used as an example! The flux of OGA to the glutamate family is 0.99 to Glu of which 0.332 goes to protein biosynthesis, 0.237 to Pro, 0.188 to Gin and 0.237 to Arg. Again the fluxes to all of these can be calculated, but Fig. 8 gives only Arg in detail. The overall equation for the biosynthesis of 1 mol Arg is:
93
taken from the metabolic pools within the cell. These metabolic pools are kept constant by regeneration. Carbamoyl (P) is made from CO 2 + 2ATP by a synthetase using Gin as amino donor which is, in turn, regenerated from Glu, NH 3 and ATP by Gin synthetase: CO 2 + 3xATP + NH 3 = Carbamoyl(P) + 3xADP + 2x(P) Acetyl CoA is regenerated from acetate, ATP and CoA by acetokinase and a specific transacetylase: Acetate + ATP + CoA = AcCoA + ADP + (P) Glu is regenerated from OGA, NADPH,H ÷ by Glu dehydrogenase:
NH 3 and
OGA + NH 3 + NADPH,H ÷ = Glu + NADP ÷ Asp is regenerated from fumarate, H20, NH 3,
Giucose6P ~
Glycerol
Monomers
Fructose6P
GlyceroP
\/
- TriosePs~
Monomers
3PGlycerate ~
Monomers
yruvae ~ "P,~t
Monomer8
Penol
M....... ~ P y r u v a t e
AcOetaale~
M. . . . . . .
OGA + NH 3 + 2xNADPH,H+ + 2xATP + AcCoA + Glu + Carbamoyl(P) + Asp = Arg + 2xNADP ÷ + ADP + AMP + 2x(P) + (P.P) + Acetate + Fumarate The inputs are OGA and CO 2, both of which are generated by the CMPs, ammonia which is a component of the medium, together with NADPH,H ÷, ATP, AcCoA, Glu, Carbamoyl (P) and Asp which are all
AcetyI.P
[~
ADP~ATPAceiso tateCitraox~o te
~ Glutarat~_~
Excrete
Monomers
Fig. 10. Flux analysis of growth of E.coli ML308 on glycerol.
94
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
NAD ÷ and NADPH,H ÷ by fumarase and malate dehydrogenase to give oxalacetate which is transaminated to Asp by Glu which, in turn, is regenerated by glutamate dehydrogenase:
tion of the CMPs: provision of precursors and the energy required to convert them into monomers and thence biomass.
Fumarate + H 2 0 + NH 3 + N A D + + NADPH,H +
2.4. Flux analysis o f growth on glycerol
= Asp + NADH,H+ + NADP + Overall the conversion of the precursor, OGA, to the monomer, Arg is according to this equation: OGA + CO 2 + 4xNH 3 + 4xNADPH,H ÷ + NAD + 6ATP = Arg + 4xNADP÷ + NADH,H ÷ + 5xADP + AMP + 5(P) + (P.P) The overall balance of the various reactions is gathered in Fig. 9 to show the dual function of the CMPs. Flux through the CMPs generates the two precursors, OGA and CO 2, required for Arg biosynthesis. In addition the CMPs provide the 'energy', reducing power and ATP, required to drive Arg biosynthesis, a large part of which is used to regenerate the pools of intermediates also fed into the process. Finally, in the catabolic role, the CMPs must regenerate the pools of ATP, NAD + and NADPH,H +. This illustrates very well the dual func-
CHO ~Ii=O CH)O(P)
.~
Glyceraldehyde3(P)
CH2OH C=Oi HO--C-H i H-C-OH i H-C-OHI CH20(P)
Growth on glycerol is slower (74%) than on glucose, and fluxes [4] reflect this (Fig. 10). Furthermore there is no acetate excretion and the sole products of growth on glycerol are biomass, CO 2 and water. This is discussed more fully later (Section 3.3.3.). Although the input of glycerol is greater on a molar basis, the delivery of carbon into the CMPs in only 66% of that on glucose. The uptake of glycerol is by facilitated diffusion and one specific enzyme (glycerokinase) followed by glycerophosphate dehydrogenase to yield C3P. From this junction, gluconeogenesis is by fructose bisphosphatase rather than the phosphofructokinase used on glucose. Otherwise routes to precursors are the same as for glucose except, of course, there is no requirement of PEP for uptake and the flux into the PPP to generate C5P and C4P is presumed to be from C3P (Fig. 11). Proportionally less CO 2 is generated by PDH and more is generated in the Krebs cycle.
CH2OH C=O -dr Trbnsketolase ~
~ ~
Fructose6(P)
-~
CHzO(P)
Glyceraldehyde3(P)
ICH20H C=O I HO--C-H I
H-C-OH
CH20(p)
Xylulose5(P)
~
Transaldolase
~
CH20(P)
"I"
Rlbose5(P)
"4"
CH20(P)
Sedoheptulose7(P)
CHO H-C-OH I H-C-OH H-C-OH ~ ' ~ CH20(p )
H - C,- O H H--C-OHm
~ F
Erythrose4(P)
CH2OH C=O HO--C-H I H-C-OH I H-C-OHm
Clio
C=O
I CH20(P )
CHOI H--C-OH , H--C-OH CH20(P)
Fructose6(P)
Transketolase
CHO 1 H-C-OH i H-C-OHI CH20(P) Efithrose4(P)
CH2OH C=O HO--~:-H t
I
CHO C=O,
~-
Xylulose5(P)
CH2OH C=O HO--C-H H - Ci- O H H-C-OH H-C-OH
HO--C:-H H - Ci- O H CH20(p )
"I"
Glyceraldehyde3(P)
H-C-OH CH20(P)
Sedoheptulose7(P)
Fig. 11. Nonoxidativeconversionof glycolyticintermediatesto C4 and C5 sugars in the PPP.
H. Holms / FEMS Microbiology Rm,iews 19 (1996) 85-116
Glucose6P"~O"~5-~ Monomers Fructose6P
TriosePs~
Monomers
3~rate ~
J
M. . . . . . . ~
excreted as acetate and 38% fluxed to citrate in the Krebs cycle and thence 0.9% into OGA (the flux to OAA is via PEP). Compared with glucose, a large amount of pyruvate is oxidised by PDH and the Krebs cycle to CO~ and a very large amount of carbon is excreted as acetate. 2.6. Flux analysis of growth on acetate
I
Pyruvate
95
Monomern
Peno~vruvate~ PepSynJ
~
~ru~ate
Monomers PEPC
.=l~te ~
M....... ~ ~ A 3 ( ~e~t~[~
M. . . . . . .
Growth on acetate is slow (/~ = 0.43) and delivery into the CMPs is by acetokinase and a phosphotransacetylase to generate Ac.CoA. This is itself a precursor for biosynthesis but all other precursors must be generated from it by flux through the glyoxylate bypass (Fig. 5). This route, unique to growth
~
T;;/
Glucose6P"~-~4--~- Monomers Fructose6P
/
I
AOPl
iso
Citrate
FructosebisP
Acetate Oxo
°'u'srate--
TriosePs~
Exc~;ete Excrete u.ttt_2zLL_~Monomers Fig. 12. Flux analysis of growth of E. coli ML308 on pyruvate.
Growth on pyruvate is as fast ( / x = 0.73) as glycerol but flux analysis [13] shows (Fig. 12) big differences between pyruvate and glycerol and glucose. The uptake of pyruvate is very large and, on a carbon basis, is 3.t times greater than glycerol and twice that for glucose. 14.2% of the pyruvate uptake is phosphorylated to PEP (by PEP synthetase) and used directly for biosynthesis, anaplerotic provision to the Krebs cycle or gluconeogenesis by reversal of glycolysis for generation of precursors for biosynthesis. Of the pyruvate taken up (100%) a small amount (I .4%) is reduced to lactate and excreted and 3.9% is used for biosynthesis. The rest (81.9%) is oxidised to Ac.CoA. Relative to the original pyruvate uptake, 4,5% of the Ac.CoA is used for biosynthesis, 39.2%
3~rate ~
Monomers
~
Monomers
PenolPyruvate
2.5. Flux analysis o[ growth on pyru•ate
M....... ~
Pyruvate~. ~
Monomers ~
Acetate
[ ~
M.......
~Oxal
A~ety[
/~
Monomers
Acetate
ICx~O~'Acetate/s°lCitr~~' e~Glutarate *~umarate~e ~
Monomers
Fig. 13. Flux analysis of growth of E. coli ML308 on acetate.
96
H. Holms / FEMS Microbiology Reuiews 19 (1996) 85-116
Glu~ose6P~
on acetate and other 2-carbon inputs, generates malate and thence O A A which flux to Pyr and PEP, respectively. On these assumptions the measured data can be calculated as a flux analysis (Fig. 13). The implications of these assumptions must be clearly defined. These are: • the flux to A c . C o A is directly from the carbon
"l'riosePa~
3~rate
Pyruvate
~
Morlorner8
Monomers
~
Monomers
PenolPyruvate---~.~-~M . . . . . . .
Oxal
P ruvate
Acetyl~
8.471
~' Glucose6P-~'-~3"~Monomers
TS•
L
Fructose6P
Glucose
uctob~se ~--]-~-----
~t = 0 . 7 2 Chemostat
Fr
L ctate
I
/ ~ ......
Monomers
j Peno~ate ~
Monomers
~Pyru~ate
AcOe~ate~
Acetate
Excrete
Fig. 15. Flux analysis of E. c.oli ML308 in a pyruvate chemostat at /x = 0.30.
M. . . . . . .
~
ADP p~ AT
Acetate 0~o - ~ Glutarate
,
3PGlycerate ~
~
~,._
AcetyI.P ADP= iSOCitrate
isP
Excrete
M....... ~ 0 - " ~
.......
Fructoae6P
source;
• there is no need for a flux from pyruvate to A c . C o A and PDH does not operate; and • the route to pyruvate is from malate by malate enzyme (ME) and to PEP is from O A A by PEP carboxykinase (PEPCK). This illustrates a fundamental problem in the technique of flux analysis. W e need to know the fluxes from precursors to monomers and the routes by which the carbon source sustains these fluxes. The
. .......
M
isoCitraox~o te Glutarate
Excrete
Fig. 14. Flux analysis of E. coli ML308 in a glucose chemostat at /z = 0.72.
flux from precursors is no problem because it is derived from a measurement of the monomers generated for biosynthesis. In this case, however, the metabolic routes employed are a matter of judgement. W e know, for certain, that there must be a flux of 1.00 to pyruvate and o f 2.41 to PEP and we choose to make these by ME and P E P C K respectively. However, in principle, at one extreme there could be a flux from O A A to PEP of 3.41 and thence, to pyruvate the flux o f 1.00 by pyruvate kinase. At the other extreme, there could be a flux of 3.41 from malate to pyruvate and thence to PEP a flux o f 2.41 by PEP synthase. In between, there could be any combination of these fluxes. W e choose the simplest, most economical solution which is an anthropocentric view of what E. coli would be best advised to adopt! However this may be, there is no
97
H. Holms / FEMS Microbiology ReL,iews 19 (1996) 85-116
doubt that the net fluxes to pyruvate and PEP must be 1.00 and 2.41! The unique features of flux analysis of growth on acetate are obvious: • flux to all precursors, except Ac.CoA, is by the glyoxylate bypass which must compete with flux through the Krebs cycle. In particular, ICL and ICDH compete for isocitrate; and all fluxes to precursors differ from those operating on glucose and indeed many other carbon sources.
Fructose
PTS
Glucose6P
+
'
[ ~
Fructose6P
FructoselP
Fructose bisP
TrlosePs~
2.7. Flux analysis of growth in chemostats and on other carbon sources
Flux analysis of growth on the other carbon sources and conditions will be presented in less
"~-0"~-]--'~" Monomers
Monomers~
Monomers
3PGlycerate ~
Monomers
Peno/P~yruvat~e
Monomers
- Pvru~a' yru]t e
Oxal Acetate
~
Monomers
= Glucose6P~ M o n o m e r s
Prot
on•sym port
Glucose 6P
I
[ ~
Fructose6P iso Citrate
Fructose bisP
ATP
Acetate
TriosePs~
Monomers Excrete
.......
3PGlyeerate~
Monomers
Pone yruvate~
Monomers
]
Acetate ~
Monomers
Fig. 17. Flux analysis of growth of E. coli ML308 on fructose.
Monomers
.......
AOPATPAcet ~ ateisoCitraox~o te Glutarate Excrete
Fig. 16. Flux analysis of growth of E. coli ML308 on glucose 6-phosphate.
detail. When uptake of glucose is restricted in a chemostat, acetate excretion is diminished and, at /x of 0.72, is zero (Fig. 14). Under these circumstances there is no flux through pyruvate kinase and the flux to PEP is used for uptake (73.2%) biosynthesis of aromatic amino acids (4.7%) and anaplerotic flux into the Krebs cycle (23.3%) which add up to rather more than 100% because the data have been 'rounded up'. The flux from glucose to PEP is insufficient and the remainder is shown to be generated by recycling pyruvate through PEP synthetase. This is necessary to balance the flux diagram but, of course, data are not sufficiently accurate to predict such a flux. It is more probable that there is no reverse flux through PEP synthetase which normally does not operate in a glucose culture. Nevertheless this illustrates a problem inherent in this form of flux analysis, or, indeed,
98
H. Holms / FEMS Microbiology Ret,iews 19 (1996)85 116
in any other method such as metabolic control analysis, both of which assume that we know the detail of the metabolic routes actually used in the system under study. Measurements of inputs and outputs derive net fluxes of substrate to product. If PEP synthetase was expressed in the glucose phenotype and if it did operate, a futile cycle would be possible and the flux analysis would not detect it. In the same way when pyruvate uptake is restricted in a chemostat, acetate excretion is diminished and at /z = 0.30 is totally abolished (Fig. 15). While growth rate falls to 41% pyruvate uptake is only 25% of that when unrestricted. Flux of precursors to monomers follow the fall in growth rate but oxidation of Ac.CoA in the Krebs cycle is actually increased, relative to /~, by 26% in the chemostat as compared to the batch. This increase in oxidative
Glucose6P ~
Glucuronate
Monomers
[ ~ Fructose6P
Mannonate :
\/
P\
PenolP~rLvate~
Monomers
.......
.......
Glucose6P ~-[-0"~'9"2-'~ Monomers Gluconate Acetyl.P
1
Aop~ArPAcetate
Fructose6P
isoCitr@e
6P-Gluconate
/"
~
\/
Monomers ~ 2 Keto 3 deoxy
TriosePs ~ /-~¢~_._~
Monomers
Excrete
Glutarat[~,.
Monomers
Fig. 19. Flux analysis of growth of E. coli ML308 on glucuronatc.
\
PenolPyruvate ~
Monomers~ '~''~-~ ' p- Y•ruvat Ve I
Monomers
Oxal ~ u Acetate f .... i ~ monomers
AOPATPAcet ~ ate iso Citr@te Glutarate
Excrete Fig. 18. Flux analysis of growth of E. coli ML308 on gluconate.
energy production presumably replaces the ATP generated by acetate excretion in the batch. In other words, restriction of pyruvate uptake permits it to be partitioned between the two functions of the CMPs (precursors and 'energy') in the most economical fashion. Glucose 6-phosphate is taken up by a proton symport but otherwise pathways are the same as for glucose (Fig. 16). Growth rate also is similar but 11% more carbon is taken up, t8% less acetate is excreted but 1.9 times as much Ac.CoA is oxidised in the Krebs cycle. Fructose is taken up by a PTS to generate fructose l-phosphate and, on the principle that the shortest and most economical routes to precursors are chosen, triose phosphate is shown as the route to C 5 and C 4 sugar phosphates by the PPP (Fig. 17). No acetate is
99
H. Holms / FEMS Microbiology RetJiews 19 (1996) 85 116
excreted and growth rate is slower than on glucose but equal to that in the glucose chemostat at which acetate excretion is also zero. To all intents and purposes, a fructose phenotype at /'/'max is metabolically equivalent to a glucose chemostat at /z0.72. However, the carbon uptake is 1.23 times greater and oxidation of Ac.CoA in the Krebs cycle 2.22 times greater than the glucose chemostat. Obviously the conversion of glucose to biomass at /~0.72 is much more efficient than fructose at the same growth rate. Flux analyses show that the glucose chemostat (Fig. 14) generates 4.69 tool A T P / k g / b (net) by glycolysis while the fructose culture (Fig. 17) makes 14.18 tool A T P / k g / h (net) as well as proportionally more reducing power. There is therefore no obvious reason that the fructose phenotype has to oxidise so much
Glucose6P- ~ - ~ - ~
M. . . . . . .
Fructose6P
TriosePs
~
Monomers
3~rate ~
Monomers
PenoIPy~vate~ Monomers~
Pyruvate [~ ~
Glucose6P----[O~]--~- Monomer s
Monomers
~
Oxal - Acetate~ 0.54~'~ M.......
Citi'ate
Malate
AcetyI,P [ ~ isoCitrate
Fructose6P
Fumarate
ATP
Acet 4 at Lactate
TriosePs ~
Monomers
Penoyruvate~
Acetate
~f Acetate
Monomers
isoCitrate Ox~o
Excrete
Fig. 20. Flux analysis of growth of
Fig. 21. Flux analysis of growth of rate.
E. coil
ML308 on oxogluta-
Monomers
Oxal ~
Monomers
Excrete
Monomers
+ 3~rate ~
LDH
Pyruvate
Gl~~r~sUccin2oglutarate
\/
Monomers
E. co/i
ML308 on lactate.
more Ac.CoA in the Krebs cycle. This problem is discussed further below (Section 3.3.4.). Gluconate (Fig. 18) is taken up and phosphorylated to 6-phosphogluconate which is a component of the PPP and presumed to be the source of pentoses and tetroses for biosynthesis. The EDP then generates equal amounts of triose phosphate and pyruvate. The consequence of this is that only 32% of the gluconate-carbon reaches PEP but a large fraction is dephosphorylated by PK and joins the very large flux from EDP to pyruvate. The consequence is a very large acetate excretion. Glucuronate (Fig. 19) is also metabolised by EDP and, although growth is slower (/x = 0.87) than on gluconate (/z = 0.90), acetate flux to excretion is 40% greater, and oxidation in the Krebs cycle is 31%
100
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
less. On the other hand, metabolism of glucuronate to 2-keto-3-deoxygluconate generates reducing power but, because the route does not pass through 6 phosphogluconate, precursors from the PPP are generated by the non-oxidative route. Lactate (Fig. 20) is obviously similar to pyruvate (Fig. 12) but uptake and growth are slower. Acetate excretion much less (17% of uptake, on a molar basis, compared with 39% for pyruvate). Here again is a pair of carbon sources with almost identical metabolic profiles. Lactate compared with pyruvate is slower in growth (82%), acetate excretion (24%) and oxidation of Ac.CoA (71%). The explanation must rest on the slower uptake (55%) of carbon source and the additional reducing power generated by oxidation of lactate to pyruvate. Growth on OGA (Fig. 21) is very slow (/z = 0.32) and no acetate is excreted. Uptake of carbon source is very slow and probably limiting (see Section Glucose6P
""~'-~'.6-5-~--~-Monomers
Fructose6P
TriosePs ~
Monomer=~
Monomers
3~rate
~
Monomers
Penol Py[~ate
~
Monomer,
Pyruvate
l
Acetate
Excrete
'~
"
~
Oxal - A~e;;te --~ 1.06 ~
/C.ta,o
M. . . . . . .
.um r=o
.t~o~$uc¢inate Glutarate ~
Monomers
Fig. 22. Flux analysis of growth of E. coil ML308 on fumarate.
3.3.2.). Oxoglutarate is the only carbon source in the set of eleven on which E. coli ML308 will not grow without a genotypic change (constitutive expression of the dicarboxylic acid permease). Fumarate (Fig. 22) on the other hand, supports a reasonable growth rate (/x = 0.63) without acetate excretion.
3. Discussion 3.1. The distribution of carbon fluxes in the CMPs The fluxes of the carbon sources through the CMPs (Table 2) show the extraordinary flexibility of the system. Where thermodynamics and pool sizes make a reaction physiologically irreversible, there is usually another enzyme which sacrifices 'energy' in order to drive the reaction in the 'opposite' direction. In this sense, probably only PDH and OGA dehydrogenase are irreversible. There is a large gap between the phosphorylated and non-phosphorylated parts of the CMPs. The pool of PEP is at the centre of this junction and the fluxes in and out cover a very wide range depending on the carbon source. Enolase, PEP synthetase and PEPCK generate PEP within the range of 7.36-18.3, 0-6.1 and 0-3.5 respectively. Fluxes out are by 'reversal' of enolase to feed gluconeogenesis (1.5-6.1), PTS for uptake (0-10.6), PK (0-15.2) and PEPC for anaplerotic provision to the Krebs cycle (0-2.6).Whatever the nature of the carbon source or special circumstances of any one experiment, the function of the CMPs is always to sustain fluxes to precursors and the energy required for biosynthesis. The fluxes to precursors are a question of measurement as are the other carbon outputscarbon dioxide and, on some carbon sources, acetate. The energy required for conversion of precursors to monomers and their assembly into biomass is generated by the CMPs as a small amount of ATP and a much larger amount of reduced nucleotides ([2H]). Some [2H] is used to make monomers more reduced than their precursors and, as we know the biosynthetic pathways this is subject to calculation. We do not know the amount of ATP made by oxidation of the balance of [2H] available but it must be at least sufficient to power biosynthesis and growth. This will be dealt with later (Section 3.3.4.). The data from the flux analyses (Fig. 7, Fig. 10,
101
H. Holms / FEMS Microbiology Reviews 19 (1996) 85- 116 Table 2 Fluxes through key enzymes with CMPs on eleven carbon sources
G6P Glucose Gluconate Glucuronate Fructose Glucose chemo Glycerol Pyruvate Fumarate Lactate Acetate Pyr chemo Oxoglutarate
Growth rate /*
PTS
PFK
FbisP ase
TPDH
Enolase
PK
PEP Syn
PDH
PEPC
PEPCK
ME
Acetate excretion
(I.95 0.94 0.90 I).87 I).72 11.72 0.70 11.73 I).63 0.60 0.43 0.35 0.32
0 10.57 11 1) 7.62 6.22 0 0 0 1) 11 11 0
9.88 8.71 0 0 0 4.79 0 11 11 0 0 0 11
0 0 11.92 0.89 0.74 11 0.72 0.75 11.65 0.62 0.44 I).36 0.34
19.63 17.29 8.65 8.54 12.58 9.48 11.41 - 2.69 - 2.33 - 2.22 - 1.58 - 1.29 - 1.20
18.34 16.01 7.42 7.36 I 1.6 8.5 10.46 - 6.08 - 3.19 - 3.04 2.17 - 1.53 - 1.64
t5.211 2.34 4.45 4.49 1.60 0 8.15 0 0 0 0 0 0
1l 1) 0 0 0 (I.I 0 6.118 0 5.01 0 2,92 1/
13.00 10.73 12.98 14.21 7.55 4.45 6.53 35.111 9.79 17.26 0 10.02 0
2.61 2.58 2.47 2.39 1.98 1.98 1.92 2.(X) 11 1.64 0 I).96 0
0 (I 0 0 0 11 11 (1 3.54 0 2.41 0 1.82
I) 0 0 11 0 0 0 I) I 1.25 0 1.00 0 3.95
4.(10 4.89 6.48 9.117 II I) 0 16.81 0 4.06 -20.16 11 0
FbisPase: fructosbisphosphatasc. TPDH: triosephosphate dehydrogenase.
Fig. 12, Fig. 13, Fig. 14, Fig. 15, Fig. 16, F i g . 17,
a l s o m a d e by o x i d a t i o n o f i n t e r m e d i a t e s to t h e car-
Fig. 18, Fig. 19, Fig. 20, Fig. 21 a n d F i g . 2 2 ) are
b o x y l i c a c i d s w h i c h are t h e s u b s t r a t e s f o r d e c a r b o x y -
a s s e m b l e d in T a b l e 3 b y g r o w t h rate. F l u x o f s o m e
lation. In a s e n s e , f l u x to c a r b o n d i o x i d e r e f l e c t s t h e
c a r b o n s o u r c e s into t h e C M P s e x c e e d s t h e f l u x e s to
p o t e n t i a l o f t h e C M P s to t r a p ' e n e r g y ' . B e this as it
carbon
is
m a y , t h e r a t i o o f t h e total c a r b o n f l u x e d to p r e c u r s o r s
e x c r e t e d as a c e t a t e . T h i s g e n e r a t e s A T P a n d p r e v e n t s
u s e d f o r b i o s y n t h e s i s to t h e flux o f c a r b o n e v o l v e d
accumulation
T h e b u l k o f t h e c a r b o n d i o x i d e is m a d e b y o x i d a t i v e
as c a r b o n d i o x i d e ( P r e - C / C O 2) tells us s o m e t h i n g quite profound about the organisation of the CMPs.
d e c a r b o x y l a t i o n ( b u t P E P C c o n s u m e s C O 2) a n d t h e s e
T h o s e f e e d s t o c k s w h i c h f e e d into t h e p h o s p h o r y l a t e d
reactions
intermediates
dioxide and precursors and the balance
of low molecular weight compounds.
generate
reduced
nucleotides
which
are
of the
CMPs
have
a much
higher
Table 3 Inputs and outputs of the CMPs on eleven carbon sources Carbon source
Condition
Growth Uptake Acetate Carbon dioxide Precursor carbon ( m o l / k g / h ) Pre-C/CO z /* (h i ) ( m o l / k g / h ) ( m o l / k g / h ) ( m o l / k g / h )
Glucose 6-phosphate Batch 0.95 Glucose Batch/turbidostat 0.94 Gluconatc Batch 0.90 Glucumnate Batch 0,87 Fructose Batch 0,72 Glucose Chemostat 0,72 Glycerol Batch 0,70 Pyruvate Batch/turbidostat I).73 Fumarate Batch 0.63 Lactate Batch 0.60 Acetate Batch 0.43 Pyruvate Chemostat 0.30 Oxoglutarate Batch 0.32 Pyruvate Batch and bromopyruvate 0.30
11.76 10.57 1 1.48 11.74 7.62 6.22 14.1X) 42.87 16.52 23.66 20.16 12.50 6,65 12.98
4.(X1 4.89 6.48 9.07 0 0 11 16.81 0 4.06 - 211.16 0 0 1)
Pre-C/C02: flux to total precursor carbon/flux to net C02 evolution.
23.27 14.78 18.66 16.58 16.11 7.8t) 13.23 64.99 40.15 38.21 22.64 25.06 20.01 26.55
40.20 39.76 38.12 35.72 29.61 30.51 28.77 29.94 25.93 24.65 17.68 12.44 13.24 12.47
1.73 2.69 2.04 2.15 1.84 3.9 I 2.17 l).4~ !).65 0.65 0.78 I).511 0.66 0.47
102
14. Holms/FEMS Microbiology Reviews 19 (1996) 85-116
P r e - C / C O 2 ratio (1.7-2.69) than those which feed into the non-phosphorylated intermediates (0.550.78). This is true of those feedstocks which flux through the EDP and divide their carbon between phosphorylated and non-phosphorylated intermediates so that the destination of half the carbon confers this advantage of high P r e - C / C O 2. Compare glycerol (Fig. 10) and pyruvate (Fig. 12). The enzymes between PEP and fructose bisP are reversible so the interconversion of PEP and pyruvate is the only metabolic difference between the two. The direction of flux (PK vs. PEP synthetase) appears to have a very profound effect on P r e - C / C O 2 (2.17 vs. 0.55) On acetate, fumarate and OGA (Fig. 13, Fig. 21 and Fig. 22) flux to PEP and gluconeogenesis is by PEP carboxykinase which appears to be equally disadvantaged ( P r e - C / C O 2 of 0.78, 0.65 and 0.54). Overall the carbon sources entering the non-phosphorylated parts of the CMPs must overcome the problem of energy generation for gluconeogenesis. The gulf between the phosphorylated and the non-phosphorylated parts of the CMPs is more easily crossed in the direction towards the carboxylic acids which are the prime substrates for oxidative metabolism. One could perhaps anticipate this on the grounds that the phosphorylated pathways were themselves sufficient to support growth for the larger part of the evolution of the CMPs. It follows that when oxygen entered the ecosystem the task of redesigning the phosphorylated CMPs to take advantage of the hugely increased availability of ATP was too complex to be selected. The consequences of this will be discussed below (Section 3.3.) however it should be stressed that this argument applies to E. coli and there are other organisms (e.g. Pseudemonas and Acinetobacter species) which actually prefer carbon sources which enter the non-phosphorylated parts of the CMPs. It would be very interesting to measure P r e - C / C O 2 ratios for these organisms on different carbon sources. 3.2. Intervention to reduce acetate excretion
One very great advantage of presenting data as flux diagrams is that they illustrate where it might be possible to intervene and change metabolic fluxes. The flux diagrams for all those feedstocks which sustain acetate excretion (Fig. 7, Fig. 12, Fig. 16,
Fig. 18, Fig. 19 and Fig. 20) show certain features in common: the uptake of carbon source and flux into the CMPs exceeds their capacity to generate precursors and convert them to monomers; the flux through PDH produces more Ac.CoA than can be used for precursors or energy generation; phosphotransacetylase and acetokinase act as a safety valve to excrete intermediates from the CMPs; and the CMPs of all acetate excretion phenotypes are divided into two parts by the conversion of PEP to OAA. It is therefore possible that PEP carboxylase limits flux to OAA and OGA in these phenotypes. These possibilities will be dealt with in turn.
3.2.1. Excessive uptake
The easiest way to restrict the uptake of a carbon source is to restrict its availability and the simplest way to do that is in a chemostat. Glucose and pyruvate both excrete acetate but feed to opposite sides of the division in the CMPs. In both cases restriction of uptake in a chemostat (Table 3) can totally abolish acetate excretion (Figs. 14 and 15). In the glucose phenotype, acetate excretion falls to zero when the growth rate is 0.72 (76.6% of /Zm~x) but glucose uptake is reduced proportionally more (to 59%). The obvious effect on the CMPs is that PK is completely redundant (Fig. 14) and PDH falls to 41%. Glycolysis and the Krebs cycle are reduced by only slightly more than uptake. On pyruvate (Fig. 15), a much greater restriction must be imposed on uptake, down to 26%, to prevent acetate excretion. Growth is lowered to 42% as is flux through PEP synthetase (PEPsyn), PEPC and glycolysis all of which have only a biosynthetic function in this phenotype. PDH falls in line with uptake (to 26%) but flux round the Krebs cycle to 52%. In both cases reduction in uptake lowers growth rate and eliminates acetate excretion. The reduction required in pyruvate uptake is much greater than glucose because the enormous inward flux is also reflected in the largest acetate excretion seen on any carbon source which, on a molar basis, is no less than 47% of uptake.
H. Holms / FEMS Microbiology Reuiews 19 (1996) 85-116
3.2.2. Excessiue flux through PDH Clearly if the flux through PDH could be reduced this would diminish the amount of Ac.CoA available for excretion as acetate. 3-Bromopyruvate is an inhibitor of PDH with a well-known mode of action [15,16] which, added (50/xM) to a pyruvate batch culture, prevents flux to acetate excretion (Table 3). Growth rate (0.30), uptake (12.98 m o l / k g / h ) are rather more inhibited than in the chemostat but otherwise the effects on fluxes are similar [ 13]. One might be tempted to say that bromopyruvate inhibits uptake of pyruvate but it is probably more true to say that unrestrained PDH activity stimulates pyruvate uptake! 3.2.3. Cont;ersion of Ac.CoA to acetate It is commonly said that excretion of acetate is the reversal of its uptake but Jones et al. have shown [17] that, at least in K strains of E. coli, the enzymes of acetate uptake/excretion are constitutive (Table 4). It seems more sensible, therefore, to consider these primarily as a safety valve for acetate excretion. Be that as it may, it is relatively easy to raise mutants deficient in the acetate excretion enzymes by enriching those insensitive to fluoroacetate [ 1 8 20]. This mutant was derived from E. coli. ML308 [13] and grown on pyruvate. Flux analysis (Fig. 23) confirmed that acetate was not excreted but that lactate excretion was now employed to redress the balance of what was still an excessive pyruvate uptake. Clearly the acetate safety valve is very important and if it is closed, the biomass must find another one in order to survive when growing on a carbon source whose uptake cannot be controlled down to a level where the CMPs are able to cope with the fluxes generated.
103
Glucose6P ---"-~O~--~Monomenl Fructose6P
\/ TriosePs
~
~
Monomer,
I 3~rate
Pyruvate M....... ~
~
Monomers
Peno~yruvate ~
Monomers
L~
yruvat'e
AcOe~aal~te
_co,. y 1
I~
M.......
[
Citi'ate /
Lactate Ace~l.P~ ] A~P'~
isoCitrate
Acetate
GlO~Oat e
Excr!leExcl~e~e~ : o ~ Fig. 23. Flux analysis of a mutant of E. coli ML308, lacking transacetylase, on pyruvate.
The transacetylase negative mutant grows on pyruvate more slowly (/z = 0.63) than its parent (/x = 0.73). However for this fall in growth rate (to 80%) the uptake of pyruvate falls to 58% (Figs. 12
Table 4 Activity of enzymes for acetate uptake in E. coli PA239 and acetate excreted in E. coli ML308 Carbon source
Acetokinase ( m o l / k g / h ) PA239
Phosphotransacetylase ( m o l / k g / h ) PA239
Acetate excretion ( m o l / k g / h ) ML308
Glucose Glycerol Gluconate Ribose Pyruvate Lactate Acetate Malate
45 39 108 39 123 111 54 51
36 39 75 36 78 84 39 45
4.9 0 6.5 16.8 4.1 - 20.2 0
104
H. Holms / FEMS Microbiology Reuiews 19 (1996) 85-116
and 23). The deviation of pyruvate flux to excretion is 39% in ML308 but only 14% in the mutant. It is quite easy to see how the block in transacetylase shifts excretion from acetate to lactate but it is not so obvious how it diminishes pyruvate uptake. Nevertheless it certainly does so! The uptake of pyruvate could be affected by a change in the proton gradient or by a feedback loop from an intracellular pool of a metabolite closely related to pyruvate metabolism (pyruvate itself or Ac.CoA?). This conclusion illustrates both the strength and weakness of the technique of flux analysis. In this case it tells you that something has controlled pyruvate uptake but gives no indication of the mechanism.
3.2.4. Enhancement of flux to OAA and OGA The flux to OAA and OGA depends on the anaplerotic flux to the Krebs cycle by PEPC in all cases where acetate is excreted (Fig. 7, Fig. 12, Fig. 16, Fig. 18, Fig. 19 and Fig. 20). This enzyme has been cloned and then inserted into a strain of E. coli K12 [21]. This strain is very different from E. coli ML308 and grows much more slowly on all carbon sources (e.g. on glucose /z is 0.47 compared to 0.94 for E. coli ML308). The only fair comparison is with E. coli ML308 where, by the same technique, the available PEPC bas been increased, by a factor of 74.6 fold, in a constructed strain, E. coli ML308/JOE4 [22]. This strain was grown on glucose and the fluxes analysed (Fig. 24). The result was unexpected! The strain grew somewhat slower (and therefore had a slower flux through PEPC) took up less glucose, diminished acetate excretion by almost 60% and had zero flux through pyruvate kinase. Interpretation of this result must await further work but the principal effect of increasing expression of PEPC is to reduce uptake of glucose by the PTS and use the glucose uptake more efficiently. Presumably the genetic intervention activates some control of the glucose PTS by changing intracellular pools of intermediates or some similar mechanism. The result is half-way to that of the glucose limited chemostat (Fig. 15) with the most obvious similarity being the abolition of flux through PK. However, whatever may be the mechanism which restricts glucose uptake, it is not as efficient as the chemostat and acetate must be excreted.
= Glucose6P "~'-~--~-Monomer$
TS•
Glucose
F
6P
Fru~c@se bisp
Monomers
3PGlycerate ~
Monomers
PenolP[~yruvat~e
Monomers
17.60 1 o,ool 1 0.031
M....... ~ P y r u ~ a t e
Acetate ~
M. . . . . . .
t,OPAT~Acet ~ ateisoCitrao~o te ~ Excrete
Glutarl~_~ Monomers
Fig. 24. Flux analysis of a recombinant of enhanced in PEPC, growingon glucose.
E. coli
ML308,
3.3. Controls and limitations in the CMPs
The prime function of the CMPs is to produce precursors and the energy required to power their conversion to monomers and biomass. The PreC / C O 2 ratios fall into two divisions (Table 3) which suggests that this combined task is more easily performed if the starting point is one of the phosphorylated intermediates of the CMPs rather than a nonphosphorylated compound and reasons have been advanced which might explain this (Section 3.1.). However flux analysis can illustrate other fundamental features of the CMPs which can best be proposed as hypotheses which are listed here and then discussed in the sections shown: • acetate excretion balances excessive inputs with fluxes to precursors (Section 3.3.1.);
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
• when uptakes balance fluxes to precursors without any need for acetate excretion, uptakes are limited by feedback control or uptake capacity (Section 3.3.2.); the unique junction created during growth on acetate is regulated in a unique way (Section 3.3.3.); flux to energy supply does not limit growth on any of the carbon sources tested (Section 3.3.4.); maintenance energy is thermodynamic inefficiency (Section 3.3.5.); the importance of junctions is not necessarily dependent on pool sizes except in the case of isocitrate in the acetate phenotype (Section 3.3.6.); 'control' of the CMPs generally lies outwith the CMPs (Section 3.3.7.). 3.3.1. Excessive flux of inputs causes acetate excretion The evidence to support this contention has already been given. Consider a high and low PreC / C O 2 input: glucose and pyruvate. When inputs are restricted in a chemostat, acetate excretion falls and can be totally eliminated (compare Figs. 14 and 15 with Figs. 7 and 12). The greater the excessive flux in the unrestrained culture, the greater the restriction of input required to eliminate acetate excre-
105
tion. Furthermore elimination of acetate excretion on pyruvate, by bromopyruvate inhibition of PDH, was dependent on a major reduction in pyruvate uptake. When acetate excretion was made impossible by genetic intervention (Fig. 23), pyruvate uptake was reduced to 58% of the unrestricted flux but was still sufficient to require an excretion of lactate. Finally, while the amplification of PEPC has consequences which are difficult to understand, the fall in acetate excretion (from 4.89 to 2.00 m o l / k g / h ) is accompanied by a fall in glucose uptake (from 10.57 to 7.60 mol/kg/h). 3.3.1.1. Conclusion. Uptake of carbon source at a rate which exceeds the capacity of CMPS to flux to precursors, is balanced by excretion of acetate as the first choice or lactate if this option has been lost by mutation. 3.3.2. Balance between input and output from the CMPs depends on control of input Both fructose and glycerol feed into the phosphorylated part of the CMPs, have high Pre-C/CO 2 ratios and do not excrete acetate but have somewhat lower growth rates than other inputs of this type which do excrete acetate (Table 3). The uptake of glycerol (Fig. 25) is controlled by feedback inhibi-
Gluc~ose6P~-~2~-
M. . . . . . .
Glycerol
ATP
Fructose6P
_~
Glyce rokinas~ "~r-~-~
[~
Feedback inhibition
I
Fructose bisP
I 0.
.721
GlyceroP dehydrogenase
GlyceroP
~ FA
~, TriosePs ~
Monomers
H2
3PGlycerate ~
Monomers
yruvate ~
Monomers
Peno
Remaining m o n o m e r s and Krebs Cycle
Fig. 25. Control of growth on glycerol.
106
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
tion [23]. Entry of glycerol into the CMPs is by glycerokinase which is sensitive to non-competitive allosteric inhibition by fructose 1,6-bisphosphate. Regulation of glycerol metabolism also protects E. coli by limiting the intraceilular pool of dihydroxyacetone phosphate which otherwise can feed a nonphosphorylated pathway leading to the synthesis of toxic methylglyoxal [23]. However, for the production of methylglyoxal to reach lethal proportions, an insensitive glycerokinase must have its effect enhanced by derepression of enzyme synthesis and reduction of catabolite repression. Indeed when the only factor is insensitive glycerokinase, growth rate is 30% faster than in the wild-type [23]. It should be stressed that these experiments used K strains of E. coli whose growth rates on glycerol (0.3-0.4) are much slower than ML308 (0.7) perhaps because K strains were chosen, in the first instance, for ease of transfer of genetic material rather than for studies of metabolism. Nevertheless, what little work has been published on glycerol metabolism in ML308 [24] shows that it works on the same principle as K strains. Fructose supports slower growth than its peers (G6P, glucose, gluconate, etc.) in the high PreC / C O 2 group. There is no direct evidence that the slower growth is attributable to restriction of fructose uptake. However continuous cultures of E. coli ML308, operating as turbidostats on mixtures of glucose and fructose, do not use fructose if sufficient glucose is present to support growth. If less glucose is available, it is all used and sufficient fructose is metabolised concurrently to maintain the higher growth rate characteristic of glucose [25]. Even when glucose supplies only 30% of the carbon input and fructose 70%, the growth is at the higher rate characteristic of growth on glucose alone. Even when fructose supplies 92.5% of the carbon input (7.5% from glucose) the growth rate is still 91% of that sustained on glucose alone. The conclusion must be that fructose can sustain faster growth than that seen in a fructose-only batch or turbidostat and therefore some mechanism must restrict fructose entry to the CMPs under these circumstances. It is very unfortunate that acetate was not measured in this work and we do not know if faster growth on fructose causes acetate excretion! Finally mannitol, in E. coli ML308, sustains a growth rate (0.94) as fast as
glucose (B. Clark, personal communication) and its metabolism is virtually identical to fructose (3) which again suggests that input of fructose is restrained. There is no evidence available on possible limitation to fumarate utilisation but growth is relatively slow (0.63). This could reflect the intrinsic difficulty in reversing glycolysis and crossing the C 4 - C 3 gap on this input or a limitation in the uptake mechanism for fumarate. Oxoglutarate is not an inducer of the dicarboxylic acid permease but is a poor substrate and will only grow on dicarboxylic acid permease constitutive mutants (W.H. Holms, unpublished observations). This could explain the slow growth on OGA (0.32) if inadequate uptake restricted carbon delivery into the CMPs. 3.3.2.1. Conclusion. Restriction of uptake and delivery into the CMPs of some carbon sources (e.g. glycerol) restricts growth rate and obviates the necessity for any acetate excretion. This may also be true of other carbon sources (e.g, fructose and OGA) but others (e.g. fumarate)may have a maximal rate of uptake which does not reach the threshold required to trigger acetate excretion. 3.3.3. Regulation of growth on acetate Acetate is delivered into the Ac.CoA pool in the CMPs by acetokinase and a transacetylase. These enzymes are constitutive (Table 4) and act as a safety valve to prevent build up of intermediates when the CMPs are overloaded. The remaining enzymes for acetate metabolism, ICL, malate synthetase and ICDH-kinase/phosphatase, are repressed except when acetate is the sole source of carbon. ICL and malate synthetase convert 1 tool OAA and 2 mol Ac.CoA into 2 tool OAA. This metabolic sequence has been called the glyoxylate bypass because glyoxylate is an intermediate unique to its operation [3]. This is the only example among our eleven different carbon sources where a junction is created within the CMPs when ICL and ICDH compete for a common substrate. The first inkling (in 1971) that this was a competition was when the reversible inactivation of ICDH was lbund to be a prerequisite for growth on acetate [26]. This mechanism allows ICL to compete with ICDH so that the partition of flux at this junction meets the requirements of fluxing carbon
107
H. Holms / FEMS Microbiology Ret'iews 19 (1996) 85-116
Table 5 Flux through ICDH relative to available enzyme activity Carbon source ICDH flux ICDHactivity Activity/ (mol/kg/h) (mol/kg/h) tlux
from acetate to the precursors required for biosynthesis (Fig. 5). The principles on which this system works have been described for E. coli ML308 [14] but are applicable to other strains and indeed to other Gram-negative rods [3]. In essence, ICL has a rather low affinity for isocitrate and requires the concentration of isocitrate in the pool to be raised to about 0.6 mM [27] to sustain flux to glyoxylate and succinate (Fig. 13). ICDH has a much greater affinity for isocitrate and must be restrained. This is achieved by a reversible phosphorylation of ICDH by a bifunctional kinase/phosphatase which is co-expressed with the other enzymes required for growth on acetate. The reversible phosphorylation is the mechanism of reversible inactivation because the phosphorylated form of the enzyme has no activity. This mechanism is the primary determinant of partition of fluxes during growth on acetate (Fig. 26). Although flux through ICDH is restricted it still makes up 68% of the flux from isocitrate with the balance (32%) going through •CL. 5,7% of the primary input to Ac.CoA is used directly for biosynthesis. It takes 3 mol of Ac.CoA to make each OGA (6.7% of input) and two to make each of pyruvate (9.9%) and O A A (31.2%). The remainder (46.6%) is oxidised in the Krebs cycle but outputs to OGA, Pyr and PEP all involve decarboxylations and, when these are taken
Glucose 3.36 Fructose 5.65 Glycerol 4.68 Gluconate 4.12 Pyruvate 11.63 Oxoglutarate 2.36 Fumarate 8,12 Acetate 9,84 Acetate 1CDH 3,36 dephosphorylated
60,7 53,0 66.2 81,2 72,8 135.6 92.6 10.0 70.2
into account, 56.2% of the original input carbon is lost as carbon dioxide and 43.8% delivered into the precursor pools. The precision of the control exerted by ICDH phosphorylation is reflected in the ratio of flux to enzyme activity measured on eight of our carbon sources, Table 5 [14]. The ratio of ICDH activity in the biomass to flux through ICDH varies from six to 58 on seven carbon sources, which cover the complete range of P r e - C / C O 2, but on acetate alone is 1.02. This is the only phenotype in which flux through ICDH utilises the full available activity of the enzyme. This means that phosphorylation of
Monomers
"~"~Gluconeogenisis
M ....... " ~ Penol Pyruvate l Monomers~
Acetyl ,i CoA
Monomers ~ " " ~ ~ ~Pyruvate
Oxal - - ~
~ ~
~l'---Acetyl.P ~ ==Citrate
~'
Acetate
ADP ATP
Acetate
7 [ ~ iso
Citrate
oa,.
=LI I,--. /
JL
Succinate
18.1 9.4 14. I 12.4 6.3 57.5 11.4 1.02 20.9
~
Fig. 26. Control of growth on acetate.
,.=P)
,,, .,o
108
H. Holms/ FEMS Microbiology Reviews 19 (1996) 85-116
ICDH regulates its activity in the cell to balance the flux of isocitrate between ICL and ICDH. Although the flux through ICDH is restricted by phosphorylation, the measured flux is between 2 - 4 times greater than that on five other carbon sources and is in the same range as fumarate and pyruvate. The fact that fluxes are so much less than the available enzyme activity, in all phenotypes except growth on acetate, indicates that the fluxes are determined by the concentration of isocitrate available in the pool rather than the enzyme activity. It follows that pool sizes must be small and none can be detected except during growth on acetate. The ICDH k i n a s e / p h o s phatase dephosphorylates ICDH when its activity is not required (e.g. on adding its end-product) or when it cannot function (anaerobiosis or cyanide poisoning). The dephosphorylated activity measured under these conditions is 70.2 m o l / k g / h . This means that, to function, 86% o f the available I C D H must be inactivated by phosphorylation to permit growth on acetate. Put another way, if 86% of the I C D H was not phosphorylated flux through ICDH would be 70.2 m o l / k g / h and a large additional amount of acetate would need to be taken up and completely oxidised. 3.3.3.1. Conclusion. Growth on acetate is limited by partition o f flux at the junction where ICL and ICDH
compete for isocitrate. This junction is unique to growth on acetate. 3.3.4. Fluxes to energy conservation in the CMPs It is quite simple to sum all the fluxes for generation of [2H] and A T P in the CMPs from any flux diagram, divide the net figure by the appropriate /z and express the results as mol [2H] or ATP per kg biomass made (Table 6). If we now assume that: the same biosynthetic pathways are always used to convert a given precursor to its monomers, irrespective of the carbon source generating the precursor; and it takes the same amount of [2H] and A T P to synthesise each monomer and assemble it into biomass. THEN it follows that it always requires the same amount o f [2H] and ATP to make 1 kg biomass from the appropriate set of precursors. The problem is that the CMPs generate different A T P / [ 2 H ] relative to the carbon sources. It follows that, in some cases at least, some [2H] must generate A T P by oxidative phosphorylation. Unfortunately we do not know how much ATP is generated by each [2H] oxidised to water in each phenotype, but, as a minimum, we might assume that each [2H] could surely generate one ATP. The sum of [2H] and, A T P generated in the CMPs is then the minimum amount of 'energy'
Table 6 Fluxes to ATP, [2H], CO2 on various carbon sources and conditions Input
Condition
p~
[2H] (mol/kg)
ATP (mol/kg0
[2H] + ATP (mol/kg)
CO2 (mol/kg/h)
Pre-C/CO 2
Glucose 6-phospate Glucose Gluconate Glucuronate Fructose Glucose Glycerol Pyruvate Fumarate Lactate Acetate Pyruvate Oxoglutarate Pyruvate
Batch Turbidostat/batch Batch Batch Batch Chemostat Batch Turbidostat/batch Batch Batch
0.95 0.94 0.90 0.87 0.72 0.72 0.70 0.73 0.63 0.60 0.43 0.30 0.32 0.30
58.48 40.95 40.13 49.53 56.16 29.38 69.2 109.5 86.43 138.75 116.74 108.6 94.94 145.2
36.24 19.34 12.52 14.11 15.90 9.00 13.57 24.00 2.51 4.67 - 34.33 6.50 17.66 8.33
94.7 60.3 52.7 63.7 72.1 38.28 82.8 133.5 88.9 143.4 82.4 115. l 112.6 153.5
23.27 14.78 18.66 16.58 16.11 7.80 13.23 65.05 40.15 38.21 22.64 25.06 20.01 26.55
1.73 2.69 2.04 2.15 1.84 3.91 2.17 0.46 0.65 0.65 0.78 0.50 0.66 0.47
Chemostat Batch Batch/BrPYR
109
H. Holms / FEMS Microbiology Retqews 19 (1996) 85-116
available from the CMPs to generate 1 kg biomass from the precursors simultaneously generated by the CMPs. The precursors must be used in the amounts calculated, because we have measured, in biomass, the monomers made from them, but not necessarily all the [2HI and ATP need be consumed because of futile cycles, 'slip' or any other thermodynamic inefficiency which oxidises [2H] or hydrolyses A T P to generate heat. Indeed if the energetic input to make 1 kg of biomass is constant, it is obvious that growth on many carbon sources is less efficient than others. A m o n g the high P r e - C / C O 2 group every culture generates more [2H] + ATP to make 1 kg biomass than does glucose in a chemostat at /~ 0.72. This must mean that the glucose chemostat is the most thermodynamically efficient of all these cultures and releases less heat than the others. Of course, it does not follow that the glucose chemostat is the most efficient system possible, only that it is the most efficient of this set. In the low P r e - C / C O 2 group several phenotypes generate less ATP than the glucose chemostat but this is because these feedstocks feed into the non-phosphorylated pathways which generate less A T P by substrate-level phosphorylation than the phosphorylated pathways. Indeed the only reason that pyruvate gives more A T P than the glucose chemostat is the A T P made available by acetate excretion. However the low P r e - C / C O 2 group, by their point of entry to the CMPs, generate very large amounts of [2H] and the net sum of [2H] plus ATP is 2 - 4 times that of the glucose chemostat. The only
conclusion can be that, of all the carbon sources and conditions tested, the glucose chemostat at /z0.72 is the most thermodynamically efficient, that is to say generates the least heat. There is yet another difference to be noted between the low and high P r e - C / C O 2 group. The glucose chemostat is more efficient than the turbidostat and growth restriction progressively raises the P r e - C / C O 2 ratio from 2.69 to 4.07 at the growth rate when acetate excretion is zero. As the input of glucose restricts /x from 0.94 to 0.72, excretion of acetate declines to zero, 23% less glucose is used to make lkg biomass, 31% less carbon dioxide is evolved, 53% less A T P and 28% less [2H] are generated in the CMPs and P r e - C / C O 2 rises from 2.69 to 3.91 (Table 7). Pyruvate gives totally different results. As /z is restricted from 0.73 to 0.30 when acetate excretion is eliminated, 30% less pyruvate is used to make 1 kg biomass, carbon dioxide evolution falls a little and then recovers, less ATP is generated but [2H] is virtually unchanged. The PreC / C O 2 index is also stable and low. There is certainly a large difference in behaviour between glucose and pyruvate. Pyruvate feeds into the non-phosphorylated part of the CMPs which contain only carboxylic acids, the substrates for decarboxylation and oxidation. It is obviously a much simpler task for the CMPs to generate precursors and energy from carbon sources which feed into the phosphorylated parts of the CMPs. W e do not know if the other members of the high P r e - C / C O 2 group would be-
Table 7 Fluxes to acetate excretion, energy and precursors on glucose and pyruvate in chemostat cultures Carbon Source
/x
Uptake (mol/kg)
Acetate (mol/kg)
CO2 (mol/kg)
ATP (mol/kg)
[2H] (rnol/kg)
Pre-C/CO 2
Glucose Glucose Glucose Glucose Glucose Pyruvate Pyruvate Pyruvate Pyruvate Pyruvate Pyruvate Pyruvate
0.94 0.88 0.82 0.81 0.72 0.73 0.70 0.68 0.61 0.47 0.37 0.30
11.24 10.66 9.99 9.64 8.64 58.8 51.2 48.0 43.9 42.0 40.9 41.3
5.20 4.05 2.71 2.26 0 23.1 15.3 11.7 6.2 2.7 0.7 0.0
15.72 14.30 13.12 11.90 10.83 89.1 86.8 79.4 78.5 79.4 80.3 82.7
19.34 16.92 14.27 12.86 9.00 24.0 16.9 13.2 9.2 7.2 6.2 6.5
40.95 36.83 34.21 32.05 29.38 109.5 104.0 100.4 101.0 103.3 105.2 108.6
2.69 2.97 3.23 3.56 3.91 0.46 0.47 0.52 0.52 0.52 0.51 0.50
110
14. Holms/FEMS Microbiology Reviews 19 (1996) 85-116
have like glucose in a chemostat or the low PreC / C O z like pyruvate but it seems probable because the difference between glucose and pyruvate simply reflects a consequence of the structure of the CMPs and the fact that they were originally selected for anaerobic growth. The fundamental problem of all metabolism is that the same pathways perform both functions but, in the aerobic mode, any amount of reducing power can, in principle, be fed into the electron transport 'sink'. 3.3.4.1. Conclusion. The CMPs function to provide precursors and energy. On all carbon sources tested in batch or turbidostat culture they generate more ATP and [2H] than is required to convert the precursors, which they also generate, into new biomass. The CMPs work more easily on carbon sources which feed the phosphorylated parts of the CMPs. 3.3.5. Maintenance energy The mathematical analysis of chemostat data suggested that a fraction of catabolism did not contribute to growth and this is certainly also true of the growth of E. coli ML308 considered here. In 1965, Pirt [28] attributed the non-growth related energy utilisation to 'maintenance' of the existing bacterial structure and suggested it was independent of growth rate. This would be a very anthropocentric view of a biological phenomenon as a machine, were it not for the fact that our experience suggests that the harder (faster) machines are driven, the more maintenance they require. In 1976, Neijssel and Tempest [29] pointed out that the data were not incompatible with linear dependence of maintenance on /z which later evolved into their theory of 'slip'. There can be no doubt that E. coli ML308 devotes progressively less ATP and [2HI to non-growth purposes as its growth rate on glucose and, to a lesser extent pyruvate, is lowered to the point where acetate excretion is eliminated. Therefore, if we accept that there is such a thing as maintenance, we would support the proposition that it is directly related to growth rate. This would seem reasonable because maintenance is the replacement of bits of the machine that no longer function properly. As the biological machine is made up of polymerised monomers, maintenance presumably requires the detection and re-synthesis of defec-
tive polymers. If these can be degraded to their monomers and recycled the only input required is energy. However E. coli was selected in successive periods of 'feast and famine' [30]. This scenario demands survival of famine and control of feast. We already know that long-term survival of prokaryotes is enhanced by enforced idleness and we routinely impose that by dehydration or very low temperatures. The closer 'maintenance' is to zero at zero growth rate, the better the chance of survival. We must therefore accept variable 'maintenance'. Once you have done this there is no need to adduce maintenance at all. Maintenance energy is nothing other than heat-why not call it thermodynamic inefficiency? 3.3.6. Turnover of pools The pool of isocitrate in E. coli ML308 growing on acetate is 1.54 m m o l / k g [27] and the flux through this pool is the combined flux through ICDH and ICL which is 14.43 m o l / k g / h (Fig. 13). Therefore the pool of isocitrate, during growth on acetate, turns over 2.6 times per second. The pool of isocitrate is too low to measure in E. coli ML308 when growing on other carbon sources [27]. Pools measured in other strains of E. coli which grow much more slowly than ML308 may be misleading [31]. However one study [32] on E. coli VPI9 is relevant because it grew rapidly on acetate, glycerol, gluconate and glucose at growth rates very similar to E. coli ML308 and reported pools of some phosphory-
Table 8 P o o l s o f C M P i n t e r m e d i a t e s in
E. coli ( m o l / k g d r y w t ) g r o w i n g
on glucose, gluconate, glycerol and acetate Intermediate
Strain
Glucose
Gluconate
Glycerol
Acetate
G6P
VP19
2.97
0.26
6.21
7.02
F6P
VPI9
2.97
0.26
2.16
0.04
FDP
VP19
23.76
(/.26
21.60
9.72
6PG
VPI9
0.65
3.19
0
0
3PG
VP19
2.03
2.40
2.97
0.86
PEP
K
0.21
-
-
-
PYR
K
0.90
-
-
-
Ac.CoA
K
0.93
-
-
-
Citrate
K
30.0
-
-
isoCitrate
ML308
< 0.05
< 0.05
< 0.05
1.54
Oxoglutarate
K
1.10
-
-
-
Malate
K
3.60
-
-
14. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
lated intermediates but not PEP. Pools of PEP and some non-phosphorylated intermediates were measured in a K strain of E. coli growing on glucose [33]. If we assume that these pools (Table 8) are similar to E. coil ML308 growing on the same carbon sources then we can get some idea of the turnover of pools on at least four carbon sources (Fig. 27, Fig. 28, Fig. 29 and Fig. 30). 6-Phosphogluconate is found in cells growing on gluconate or glucose by not acetate or glycerol [33]. On carbon sources which generate G6P by gluconeogenesis (e.g. acetate and glycerol) the flux diagrams assume that flux into the PPP to generate C 4 and C s sugars is by the anaerobic pathway (Fig. 11) from triose phosphates and no 6-phosphogluconate is detected in
~ Glucose6P--'--~-~- M. . . . . . . F ~ 6P~ A TGlycerol ~ ADP "~ ~. GlyceroP <~1~ \/
" TriosePs~
~
.......
M. . . . . . .
3PGlycerate ~
Monomers
Pyruva ~t e~
Monomers
Penol
Monomers
1~Glucose6P~ M
111
Pyru~ate
Oxal ~
Acetate
M....... "~--1~-~~ ~
Monomers
/
18~FFFrur~0~6i P: Acetate TriosePs ~
Monomers
4~ 3PGl~rate"~'~5~--~-M...... PenolPyruvate ~
~ ACcetl ~ " ~
M. . . . . . . ~
Acetate
GlOxo utarate [
~
Excrete
Fig. 28. Turnover of pools in E. coil ML308 growing on glycerol.
Monomers
~ / AcOtale
T |
Ox~o Glutarate
~
=.~Succinate
Monomers
Fig. 27. Turnover of pools in E. coli ME308 growing on acetate. Numbers in open circular arrows are pool turnovers (times/min).
these cultures. Where the flux diagram assumes that the PPP is fed by the oxidative pathway via 6-phosphogluconate (e.g. gluconate and glucose) this intermediate is present in the biomass (Table 8). On all carbon sources, except acetate, the isocitrate pool is undetectable and the figures given are minimum rates which could be very much bigger. When acetate is the carbon source, flux (Fig. 27) through the isocitrate pool is much faster (14.43 m o l / k g / h ) than on the other carbon sources (3.364.68 m o l / k g / h ) but the turnover of the pool is much slower ( 1 5 6 / m i n ) than on the others ( > l 1 2 2 / m i n to > 1560/min). This results from the need for a high substrate concentration to make ICL sustain the necessary flux to glyoxylate and also requires the inactivation by phosphorylation of 94.5% of the ICDH (Table 5) to restrain oxidative decar-
112
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
boxylation of isocitrate. The flux from glucose to PEP (Fig. 30) is successfully partitioned into four fluxes of which two are biosynthetic (3.2 and 16.1%) one catabolic (14.6%) and transport (66.0%). This very important junction is served by a very small pool which must be completely used and regenerated every 47 ms! The importance of any junction is not reflected in the size of the pool. Apart from fructose 6-phosphate in the acetate phenotype, none of the known pools used in gluconeogenesis (Figs. 28 and 29) is very small but there is no information for PEP available. Finally, the data on pool sizes are old and the analysis of the CMPs (both by flux analysis and others) would benefit from a complete survey of all
~_~FructosebisP TriosePs
~
127~Penot
Monomers
Pyruvate ~
Monomers
\PEP°
M....... ~ - - ~ 8 " ] ~ 3 ~
Ac~e/.~a~e "~-~
ate
Glucose6P~-[-'~-~,'- M. . . . . . . nomers
ate 10~:;TG ~ I~cOnate /
:
Monomers
+
Monomers
2Keto3deoxy /.4~.__~ gluconate 6P / / I 8,65I ~ 4 ~ 3'PGlycerate"mr'~'~-~ M. . . . . . . venoI
yruvate ~
iso Citrate
>11~
.tce~fe GlutaratOXTo e3~__
\/
\
Citrate~
AOp~ "
~ 1 ~ TriosePs ~
M. . . . . . .
.......
Fig. 30. Turnoverof pools in E. coli ML308growingon glucose.
Monomers
the intermediates of the CMPs growing on a number of carbon sources ! Monomers
Monomers
3.3.7. Control o f the CMPs ......
+ Excrete
Monomers
Fig. 29. Turnover of pools in E. coli ML308 growing on gluconate.
The function of the CMPs is to provide the precursors and energy for growth. The amount of each precursor is dictated by the monomeric composition of the biomass which, in turn, is dictated by the fraction of the DNA expressed in any phenotype. Although the composition of the cell varies with growth rate, this is most obvious when very fast growth rates are obtained in nutritionally rich media in which E. coli does not have to make many monomers de novo. We assume that when E. coli ML308 is grown on any of the eleven carbon sources, and it makes all its monomers de novo from that
14. Holms / FEMS Microbiology Ret,iews 19 (1996) 85-116
source, that the provision of precursors does not require to be varied over the relatively small range of /x involved. Further, we assume that the biosynthetic pathway from each precursor to each monomer is invariable and the amount of energy required to drive these, routes and then assemble the monomers into biomass is also constant. The biosynthesis of monomers and their polymerisation is very tightly controlled because we have never found any quantities of surplus materials accumulating in the cultures. It follows that the flux to precursors is exactly matched to the biosynthetic requirements. Lastly we conclude, from some evidence (Section 3.3.4.), that energy supply does not limit the processing of precursors into biomass and that a fraction of the potential 'energy' generated by flux through the CMPs is dissipated as heat. Why has some simple control mechanism not been selected to improve thermodynamic efficiency? The basic pattern of the CMPs was selected to cope with anaerobic metabolism when the provision of energy was probably limiting. When an E. coli ancestor acquired the ability to use [2H] to reduce molecular oxygen it increased its potential for ATP generation by an order of magnitude. However, to use all this energy for biomass generation it would have had to increase the flux to each precursor, in exactly the correct proportion, to take full advantage of the opportunity and this is the limitation. The provision of the correct mix of precursors is limited by the prehistory of the machine devised to make them which evolved for anaerobic metabolism. The relative fluxes from the precursors are fixed relative to each other by the spectrum of monomers required to generate biomass and this is dictated by the genome. It is tempting to say that the inability to sustain these fluxes aerobically without generating more 'energy' than is required to utilise them for growth is the fundamental reason for the thermodynamic inefficiency observed. However this is a hypothesis which can only be tested by further experiment. Because flux to all precursors would require to be changed simultaneously, the entire structure of the CMPs would require modification and this is clearly a difficult objective for selection. E. coli ML308 is forced to dispose of its excess [2H] by reducing something as it has always done throughout its evolutionary history. The conclusion
113
must be that, if there is any limitation within the CMPs, it must be to the supply of the complete range of all the precursors required for biosynthesis. In the simple overview of flux from carbon source to new biomass (Fig. i) the fluxes from CMPs to precursors must contain the flux which limits growth rate. Growth rate of each phenotype is probably limited by one unique flux to one precursor. Relief of this limitation would shift the limitation to the flux of another precursor, Jensen and Pedersen [34] considered the polymerase reactions which generate polymers and concluded that these were "subsaturated with precursors and catalytic components" and thus restricted growth rate. This conclusion is consistent with the view expressed here and explains how polymerisation reactions are controlled to cope with limitation in supply of precursors from the CMPs. The CMPs are not controlled to respond to the maximum potential demands of biosynthesis of monomers or the polymerase reactions for which they are the substrates. 3.3.8. Control theories Flux analysis describes the metabolic activity of the biomass, in particular the distribution of fluxes at metabolic junctions. It does not describe how these are controlled. There is no shortage of speculation to answer this question. Basically there are two theories which are usually presented in opposition to each other and have generated a vast literature which can best be summarised by two short reviews [35,36] published simultaneously. Kacser [35] has proposed a mathematical model of control theory in which the basic assumption is that there is a finite amount of control in any system and the sum of the control coefficients of all the enzymes in any system is defined as 1 (100%). When the effects of very small perturbations in steady states are examined experimentally, the conclusion is that 'control', defined in this way, is distributed throughout the system and it is rare to find any one enzyme reaction with a very high control coefficient. Even if a reaction with a control coefficient close to 1 is found it does not follow that this reaction is the 'pacemaker' for the whole system because the theory includes negative control coefficients. The significance of these cannot be overemphasised. If, for example, we are looking at control of flux to a product we might find that one
114
14. Holms / FEMS Microbiology Reciews 19 (1996) 85-116
enzyme on the route has a high control coefficient and we might be tempted to conclude that this is significant for control of flux to product. However if 'upstream' from this there is a junction at which flux is diverted to a product (excreted or not) that we have not yet discovered then that flux could well be the determinant, by a large negative control coefficient, of flux to product. The conclusion is obvious. If you do not know everything that the system is doing, you cannot start to work out how it is controlled. The solution is also obvious. You must analyse flux, that is measure everything that goes in and everything that comes out, before you can consider how the system is controlled. Newsholme [36] has called attention to the powerful allosteric controls exerted on irreversible enzymes in the CMP's. The supporters of Kacser's control theory have pointed out that, in the steady state, these enzymes do not seem to exert any degree of control much greater than other enzymes. It seems to me that both sides in the confrontation offer experimentally verifiable facts and it would be more sensible to reconcile them rather than express them in opposition to each other and I think this can be done very simply. In the steady state, fluxes through all pools and the sizes of these pools remain constant. For example, the excretion of acetate when E. coli is growing on glucose is a very good mechanism to achieve this. There is no need for any one enzyme in the whole system to be rigidly controlled and that is what is observed. In other words, the flux through glycolysis to precursors and to PEP is balanced with the subsequent flux to the non-phosphorylated precursors by excretion of acetate. Pyruvate dehydrogenase is the fulcrum of this balance. If acetate excretion has a large negative control coefficient it means that the glycolytic system itself does not require much control elsewhere. However a sudden perturbation to this steady state (such as shutting off supply of an input like carbon source or oxygen) could be catastrophic. If glycolysis continued (e.g. without oxidative generation of ATP), then the ATP pool would be depleted by generation of phosphorylated intermediates. When oxygen again became available, the biomass would not be able to resume normal activity because of unavailability of ATP and the biomass would be 'dead'. The intervention of allosteric control of irreversible enzymes (in this case,
fructose bisphosphate kinase) prevents this and this type of control has evolved to protect the pools of essential cofactors (ATP/ADP; NADPH,H+/NADP ÷) from sudden perturbations in the flow of metabolites. In the steady state there is no need for these emergency controls and the pool sizes of their effectors are such that these controls are not operative. Both Kacser and Newsholme are 'right' in the sense that their theories are valid explanations of metabolic control. These theories are not mutually exclusive but the mechanisms operate in different circumstances.
3.3.9. Deductions from .flux analysis E. coli ML308 grows over a 2.2-fold range of growth rates (0.43-0.95) on ten substrates or a three-fold range if you include oxoglutarate, the uptake of which is severely limited. Flux analysis suggests that energy supply does not limit growth rate and it follows that flux to precursors must be the limitation. When the carbon taken into the CMPs exceeds their ability to flux all of it to precursors, the excess is excreted as acetate. All of these phenotypes have ample energy supply but they must flux to precursors at different rates because they grow at different rates. G6P, glucose, gluconate, glucuronate, pyruvate and lactate all have enough carbon (because they excrete acetate) but they support a range of growth rates. It follows that the slower the growth, the slower the flux to precursors. All the precursors are required in defined amounts. If only one is limiting, the utilisation of all the others must he regulated down to the proportionate level to match the limitation. This must be one of the principle functions of the CMPs and those in E. coli ML308 certainly perform this task very well because we see no sign of oversupply to any precursor (except Ac.CoA). It is extremely probable that, in any one acetate excreting phenotype, flux to one precursor limits growth and flux to the others match this. If the flux to the limiting precursor could be increased, the flux to all the others would increase by the same proportion, at the expense of acetate excretion, and the growth rate would increase. An alternative experimental approach would be to supply the monomers derived from the limiting precursor which, if they were taken up and incorporated into polymers, would
H. Holms / FEMS Microbiology Reviews 19 (1996) 85 - 116
have the same effect-less flux to acetate and faster growth. However, the problem in sustaining flux to any one precursor must also reflect the metabolic problems to be overcome in fluxing carbon input to precursor. It must be 'easier' to make G6P from glucose than pyruvate and it follows that different phenotypes will have different limiting precursors. Experiments in which families of monomers derived from one precursor were fed to different phenotypes, in the steady state, would presumably test this hypothesis. Growth, utilisation of inputs and acetate excretion would all have to be measured. Satisfaction of one limitation in flux to precursor would probably create another. Presumably metabolic control analysis would produce similar results but it must be easier to add a family of monomers than manipulate enzyme amounts. In non-acetate excreting phenotypes the results of this type of supplementation would depend on whether (or how) the uptake of carbon source was regulated and the results would certainly illuminate these uncertainties. Lastly, the systems operating in E. coli ML308 were selected long ago under conditions vastly different from those that we impose upon it in the laboratory. It is generally accepted that 'feast and famine' [30] were the conditions in which the CMPs were selected. The three principal strategies which we observe today are control of uptake, excretion of surplus carbon and thermodynamic inefficiency. Control of uptake and excretion guard against the sudden onset of feast. The disadvantages they show in the laboratory are less than maximal achievable growth rate and profligacy with abundant carbon source. Presumably conditions of 'feast' were so rare that these were infrequent disadvantages and of such short duration that there was little pressure to select against them. The CMPs were originally selected to give precursors and energy under anaerobic conditions. Later, when oxygen became available, the system previously selected for anaerobiosis was suddenly gifted a new mechanism of oxidising reduced cofactors by a superabundant inorganic gas rather than an array of scarce organic compounds. Indeed the generation of oxygen also increased enormously the amount of organic material available. Selection under anaerobiosis was forced to balance carefully the metabolism of scarce resources among fluxes to
l 15
precursors and monomers, on the one hand, with those reactions trapping the energy required to convert monomers into biomass. Ability to access oxygen meant that one part of this balancing act (energy provision) was no longer a problem. So what we see today is a system to generate precursors which must dispose of excess reducing power by oxidation which generates heat. Growth on acetate is based on a uniquely different strategy in which competition between two enzymes is resolved by a unique mechanism. Presumably this is a later addition to the CMPs but it may be the vestige of an earlier glyoxylate oxidising cycle which was replaced by the Krebs cycle [37].
Acknowledgements I am grateful to the large number of colleagues who did the experimental work on E. coli ML308 and for their patience in discussion. However, this paper is dedicated to two friends who have died: Henry Kacser and Martin Smellie. Henry and I rarely agreed because we started with different attitudes but we always accepted each other's viewpoint and knew that our ideas would eventually be subsumed in a greater vision. In the meantime, we both enjoyed argument together. Martin supported and encouraged me, as he did for so many others. Without him I would have given up and certainly not extended the principles of flux analysis into the real world of fermentation biotechnology.
References [l] Mandelstam, J. and McQuillen, K. (Eds.) (1982) Biochemistry of Bacterial Growth (3rd Edn.) Blackwell Scientific Publications, Oxford. [2] Neidhardt, F.C., Ingraham, J.L. and Schaechter, M. (1990) Physiology of the Bacterial Cell. A Molecular Approach. Sinauer Associates, Inc., Sunderland, MA. [3] Neidhardt, F.C., lngraham, J.L. Low, K.B., Magasanik, B., Schzaechter, M. and Umbarger, H.E. (1987) Escherichia coil and Salmonella ~phimurium. Cellular and Molecular Biology, Vol. 1. Am. Soc. Microbiol., Washington, D.C. [4] Holms, W.H. (1986) The central metabolic pathways of Escherichia coli: Relationship between flux and control at a branch point, efficiency of conversion to biomass, and excretion of acetate. Curr. Top. Cellul. Regul. 28, 69-105.
116
H. H o l m s / F E M S Microbiology Reuiews 19 (1996) 85-116
[5] Dawes, E.A. and Senior, P.J. (1973) The role and regulation of energy reserve polymers in microorganisms. Adv. Microb. Physiol. 10, 135-266. [6] Dietzler, D.N., Leckie, M.P., Sternheim, W.L., Ungar, J.M., Crimmins, D.L. and Lewis, J.W. (1979) Regulation of glucose utilization in Escherichia coli during maintenance of the energy charge. J. Biol. Chem. 254, 8276-8287. [7] Neijssel, O.M. and Tempest, D.W. (1979) The physiology of metabolite overproduction. Symp. Soc. Gen. Microbiol. 29, 53-82. [8] Stouthamer, A.H. (1979) Microbial Biochemistry (Quayle, J.R., Ed.), pp. 1-47. University Park Press, Baltimore. [9] Roels, J.A. (1980) Application of macroscopic principle to microbial metabolism. Biotechnol. Bioengineer. 22, 24572514. [10] Tempest, D.W. and Neijssel, O.M. (1984) The status of ATP and maintenance energy as biologically interpretable phenomena. Ann. Rev. Microbiol. 38, 459-486. [11] Demain, A.L. (1972) Cellular and environmental factors affecting the synthesis and excretion of metabolites. J. Appl. Chem. Biotechnol. 22, 345-362. [12] Meyer, H.P., Leist, C. and Fiechter, A. (1984) Acetate formation in continuous culture of Escherichia coli K12 DI on defined and complex media. J. Biotechnol. 1,355-358. [13] E1-Mansi, E.M.T. and Holms, W.H. (1989) Control of carbon flux to acetate excretion during growth of Escherichia coli in batch and continuous cultures. J. Gen. Microbiol. 135, 2875-2883, [14] Holms, W.H. (1987) Control of flux through the citric acid cycle and the glyoxylate bypass in Escherichia coli. Biochem. Soc. Symp. 54, 17-31. [15] Bisswagner, H. (1981) Substrate specificity of the pyruvate dehydrogenase complex from Escherichia coli. J. Biol. Chem. 256, 815-822. [16] Lowe, P.N. and Perham, R.N. (1984) Bromopyruvate as an active-site-directed inhibitor of the pyruvate dehydrogenase multienzyme complex from Escherichia coli. Biochemsitry 23, 91-97. [17] Brown, T.D.K., Jones-Mortimer, M.C. and Kornberg, H.L. (1977) The enzymic interconversion of acetate and acetylcoenzyme A in Escherichia coli. J. Gen. Microbiol. 11)2, 327-336. [18] Brown, T.D.K., Pereira, C.R.S. and Stormer, F.C. (1972) Studies of the acetate kinase-phosphotransacetylase and the butanediol- forming systems in Aerobacter aerogenes. J. Bacteriol. 112, 1106-1111. [19] Guest, J.R. (1979) Anaerobic growth of Escherichia coli KI2 with fumarate as terminal electron acceptor: genetic studies with menaquinone and fluoroacetate-resistant mutants. J. Gen. Microbiol. 115, 259-271. [20] Levine, S.M. Ardeshir, F. and Ames, G.F.L. (1980) Isolation and characterization of acetate kinase and phosphotransacetylase mutants of Escherichia coli and Salmonella typhimurium. J. Bacteriol. 143, 1081-1085.
[21] Chao, Y.-P. and Liao, J.C. (1993) Alteration of growth yield by overexpression of phosphoenolpyruvate carboxylase and phosphoenolpyruvate carboxykinase in Escherichia coli. Appl. Environ. Microbiol. 59, 4261-4265. [22] Okungbowa, J. (1991) Genetic manipulation of metabolic fluxes in Escherichia coli. Ph.D Thesis, University of Glasgow. [23] Lin, E.C.C. (1976) Glycerol dissimilation and its regulation in bacteria. Ann. Rev. Microbiol. 30, 535-578. [24] Edgar, W., Forrest, I.S., Holms, W.H. and Jasani, B. (1972) The control of glycerol utilization by glucose metabolism. Biochem. J. 127, 59. [25] Clark, B. and Holms, W.H. (1976) Control of the sequential utilization of glucose and fructose of Escherichia coli. J. Gen. Microbiol. 95, 191-210. [26] Holms, W.H. and Bennett, P.M. (1971) Regulation of isocitrate dehydrogenase activity in Escherichia coli on adaptation to acetate. J. Gen. Microbiol. 65, 57-68. [27] EI-Mansi, E.M.T., Nimmo, H.G. and Holms, W.H. (1985) The role of isocitrate in control of the phosphorylation of isocitrate dehydrogenase in Escherichia coli ML308. FEBS Lett. 183, 251-255. [28] Pirt, S.J. (1965) The maintenance energy of bacteria in growing cultures. Proc. Roy. Soc. B. 163, 224-231. [29] Neijssel, O.M. and Tempest, D.W. (1976) Bioenergetic aspects of aerobic growth of Klebsiella aerogenes NCTC 418 in carbon-limited and carbon-sufficient chemostat culture. Arch. Microbiol. 107, 215-221. [30] Koch, A.L. (1971) The adaptive responses of Escherichia coli to a feast and famine existence. Adv. Microb. Physiol. 6, 147-217. [31] Holms, W.H., Hamilton, I.D. and Robertson, A.G. (1972) The rate of turnover of the adenosine triphosphate pool of Escherichia coli growing aerobically in simple defined media. Arch. Mikrobiol. 83, 95-109. [32] Moses, V. and Sharp, P.B. (1972) Intermediary metabolite levels in Eseherichia coli. J. Gen. Microbiol. 71, 181-190. [33] Lowry, O.H., Carter, J., Ward, J.B. and Glaser, L. (1971) The effect of carbon and nitrogen sources on the level of metabolic intermediates in Escherichia coli. J. Biol. Chem. 246, 6511-6521. [34] Jensen, K.F. and Pedersen, S. (1990) Metabolic growth rate control in Escherichia coli may be a consequence of subsaturation of the macromolecular biosynthetic apparatus with substrates and catalytic components. Microbiol. Rev. 54, 89-100. [35] Kacser, H. and Porteus, J.W. (1987) Control of metabolism: what do we have to measure? TIBS 12, 5. [36] Crabtree, B and Newsholme, E.A. (1987) A systematic approach to describing and analysing metabolic control systems. TIBS 12, 4. [37] Holms, W.H., (1986) Evolution of the glyoxylate bypass in Escherichia coli-an hypothesis which suggests an alternative to the Krebs cycle. FEMS Microbiol. Lett. 34, 123-127.
MICROBIOLOGY REVIEWS FEMS Microbiology Reviews 19 (1996) 85-116
Flux analysis and control of the central metabolic pathways in Escherichia coli Harry Holms ~ Robertson Institute of Biotechnology, 54 Dumbarton Road, Glasgow GI 1 6AQ, UK
Received 15 April 1996; revised 18 July 1996; accepted 19 July 1996
Abstract
The growth of the bacterial cell involves the co-ordination of the fluxes of carbon into a considerable diversity of products that are the components of the cell. Fortunately the monomers from which the cell's polymers are made are themselves synthesised from a relatively small group of precursors that are the products of the central metabolic pathways. This simplification renders cell metabolism accessible to flux analysis, a method for handling experimental data to derive metabolic fluxes. Through such analysis of the growth of Escherichia coli ML308 on 11 single carbon sources in batch, turbidostat or chemostat culture general patterns are discernible. Most significant among these are that growth on different carbon sources is achieved without any obvious enzyme acting as a regulator of metabolic flux, except when acetate is the sole source of carbon. In this case a junction is created at which isocitrate dehydrogenase (ICDH) and isocitrate lyase (ICL) compete for their common substrate and this competition is resolved by partial inactivation of ICDH to match flux through ICL and this balance limits growth rate. In this sense, flux through ICDH and ICL is 'rate-limiting'. Uptake of six of the remaining carbon inputs exceeds the capacity of the central metabolic pathways (CMPs) to sustain flux to the precursors required for growth and the CMPs are balanced by excretion of acetate. Restriction of carbon uptake by chemostat progressively diminishes growth rate and acetate excretion until acetate excretion is prevented. For the four remaining carbon sources, uptake is apparently restricted and the products are biomass, carbon dioxide and water. Carbon sources feeding the phosphorylated parts of the CMPs flux relatively more carbon to precursors (Pre-C) than CO 2 when compared with carbon sources which teed into the non-phosphorylated pathways. P r e - C / C O 2 ratios for the former are 1.73-3.91 and for the latter are 0.46-0.78. Flux analysis of all 11 carbon sources shows that there is an overabundant supply of 'energy' (ATP + [2H]), generated by the CMPs, in all phenotypes and conditions down to a glucose chemostat at /x of 0.72. This excess energy is a thermodynamic inefficiency which must be dissipated as heat. E. coli ML308 probably evolved in circumstances of 'feast" and 'famine'. The two strategies selected (excretion of surplus carbon and restriction of /~) would appear to be defences against 'feast'. Presumably there are defences against 'famine'. These are not made obvious by flux
Abbreviations: Ac.CoA, Acetyl CoA; C3P, Triose phosphate; C4P, Tetrose phosphate; C5P, Pentose phosphate; EDP, Entner-Doudoroff pathway; ETS, Electron transport system; G6P, Glucose 6-phosphate; ICL, isocitrate lyase; ICDH, isocitrate dehydrogenase; ME, Malate enzyme; OAA, Oxalacetate; OGA, Oxoglutarate; PDH, Pyruvate dehydrogenase; PEP, Phosphoenolpyruvate; PEPC, Phosphoenolpyruvate carboxylase; PEPsyn, PEP synthetase; PK, Pyruvate kinase; PPP, Pentose phosphate pathway; PTS, Phosphotransferase system; MDH (de), Malate dehydrogenase (decarboxylating); PEPCK, Phosphoenolpyruvate carboxykinase; PG, Phosphoglycerate; Pyr., Pyruvate * Corresponding author. Tel: +44 (141) 330 5264; Fax: +44 (1411 330 4629. (1168-6445/96/$32.00 Copyright © 1996 Federation of European Microbiological Societies. Published by Elsevier Science B.V. PII S01 68-6445(96)(/(1(126-5
86
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
a n a l y s i s but allosteric control o f irreversible e n z y m e s w o u l d protect pools o f essential nutrients f r o m rapid depletion on the sudden onset of 'famine'.
Keywords: Flux analysis; Central metabolic pathway; Control; Escherichia coli
Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
87
2. M e t h o d o l o g y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Definitions and units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. F r o m inputs to precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Flux analysis o f g r o w t h on g l u c o s e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Flux analysis o f g r o w t h on glycerol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Flux analysis o f g r o w t h on p y r u v a t e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Flux analysis o f g r o w t h on acetate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7. Flux analysis o f g r o w t h in c h e m o s t a t s and on other carbon s o u r c e s . . . . . . . . . . . . . . . . . . . . . . . . . . .
88 88 89 9I 94 95 95 97
3. D i s c u s s i o n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. T h e distribution o f carbon fluxes in the C M P s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Intervention to reduce acetate excretion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. E x c e s s i v e uptake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. E x c e s s i v e flux t h r o u g h P D H . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. C o n v e r s i o n o f A c . C o A to acetate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.4. E n h a n c e m e n t o f flux to O A A and O G A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Controls and limitations in the C M P s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1. E x c e s s i v e flux o f inputs c a u s e s acetate excretion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2. B a l a n c e b e t w e e n input a n d output f r o m the C M P s d e p e n d s on control o f input . . . . . . . . . . . . . . . . 3.3.3. R e g u l a t i o n o f g r o w t h on acetate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4. F l u x e s to e n e r g y c o n s e r v a t i o n in the C M P s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.5. M a i n t e n a n c e e n e r g y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.6. T u r n o v e r o f pools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.7. Control o f the C M P s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.8. Control theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.9. D e d u c t i o n s f r o m flux analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
100 100 102 102 103 103 104 104 105 105 106 108 110 1 I0 I 12 113 114
Acknowledgement .............................................................
115
References
115
.................................................................
87
H. Holms / FEMS Microbiology Reviews 19 (1996) 85- I 16
1. Introduction
This review describes the fluxes in Escherichia coli growing on a variety of single carbon sources in batch, turbidostat or chemostat cultures. Virtually all the data refer to one particular strain (ML308) from the collection of the late Jaques Monod which has been deposited as ATCC 15224. It is constitutive for /3-galactosidase by virtue of a mutation in the repressor gene but otherwise presumed to be wild-type. Flux analysis is a description of the metabolic events which convert feedstocks to biomass and by-products and is derived from a precise measurement of utilisation of inputs, generation of biomass and excretion of by-products. This review will only consider carbon fluxes but fluxes of nitrogen, sulfur, phosphate, etc. can be defined in the same way. In principle, the conversion of a carbon source to biomass is simple (Fig. l). Each individual carbon source is taken up by the cell and fed into the CMPs. The organisation of CMPs is covered in many standard texts, of which some [1-3] are more userfriendly than others. The CMPs contain some 30 or so compounds all of which are phosphorylated or are carboxylic acids, several are both and two are thiol esters of Coenzyme A. The interconversions of intermediates are catalysed by enzymes organised in pathways which are glycolysis (Emden-Meyerhof pathway: EMP), the pentose phosphate pathway (PPP), pyruvate dehydrogenase (PDH), phosphoenolpyruvate carboxylase (PEPC) and the Krebs cycle or variations on these (e.g. Entner-Doudoroff
Carbon Source
pathway (EDP), phosphoketolase pathway) depending on the chemical nature of the carbon input Which dictates the point of entry into the CMPs. The number of enzymes involved is larger than the number of substrate to product conversions because some reactions have more than one enzyme available and many reactions use different enzymes depending on which direction they operate (but many reactions are freely reversible). Many reactions involve cofactors (NAD(P)+/NAD(P)H, H +, F A D / F A D H , A T P / A D P / A M P , C o A / A c . CoA and inorganic ions such as Mn 2+ and phosphate) and, in any steady state, these are recycled and maintained in constant pools. It is thus possible to consider fluxes of carbon in isolation. Glucose is taken as the standard to which other inputs are compared and feeds CMPs consisting of EMP, PPP and the Krebs cycle. A fraction of the intermediates of the CMPs are the precursors for biosynthesis and each precursor is the starting point for synthesis of monomers which, in turn, are polymerised and then organised into the polymeric structures of the biomass. These anabolic processes require energy input in the form of ATP and reducing power [2H]. These pools of 'high-energy' compounds are maintained by the CMPs which thus perform a dual (amphiboli~) function. The catabolic components of the CMPs are oxidative decarboxylation of pyruvate (Pyr), isocitrate, OGA and 6-phosphogluconate to generate carbon dioxide. The oxygen input is water with the result that there is a much greater generation of [2H] than carbon dioxide. Molecular oxygen is not utilised in the CMPs although they are able to oxidise carbon sources completely. Thus for glucose, total oxidation follows the equation: C6HI206 +
H20 Heat
)~
entr , Metabolic Pathways
I~ATP
tile cycle~PiOH R Precursors PiOH~~ X'OP~ADP~ ~eat !m Acetate H20 ~X=OJ ~ ATP.~[ Mon/ ers 2 H~O Polymers ~v Siomass
Fig. I. Functions of the central metabolic pathways (CMPs).
6H~O-~
6C02 +
1212H]
The reducing power thus generated is then oxidised in the electron transport system for which the terminal electron acceptor is molecular oxygen. The pools of reduced cofactors are kept constant by flux of electrons (and associated protons) through the electron transport system (ETS) in which the terminal electron acceptor is molecular oxygen. The flow of electrons down the thermodynamic gradient of the ETS is coupled, by the proton motive force, to the phosphorylation of ADP to ATP.
88
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
The uptake of glucose into the CMPs must satisfy all these fluxes derived from the CMPs. The fundamental requirement asked of the CMPs is the provision of precursors in exactly the amounts required to sustain fluxes to monomers which, in turn, is specified by the DNA and the flux to polymers defined by the expression of that fraction of the DNA specific to the glucose phenotype. In other words, the relative fluxes from precursors are predetermined and their conversion to biomass necessarily also determines the demands for ATP and [2H] for their anabolism. As a minimum, the uptake of carbon source must satisfy these demands and, if it does this, it either matches them precisely or limits their rate, i.e. limits growth rate (tz). However, in the particular case of glucose, uptake delivers carbon into the CMPs at a rate which exceeds the requirements of flux to precursors and energy. The total pool of solutes in E. coli is constant [4] and the cell wall is designed to contain the consequent osmotic pressure but it could not sustain the much higher pressures which would result from accumulation of excess carbon as low molecular weight compounds. It follows that excessive uptake of carbon source must be accommodated by other mechanisms of which there are three: intermediates in excess of requirements can be converted to storage polymers [5,6]; the coupling of ETS to phosphorylation may operate at less than maximal thermodynamic efficiency [7-10]; and the surplus carbon fluxing through the CMPs can be excreted as low molecular weight compounds [4,11-13]. E. coli ML308 does not accumulate polymers when flux to amino donors is sufficient and, as is shown below, some excess carbon is oxidised to generate heat and the remainder is excreted, usually as acetate. How can the fluxes of this system, and others, be measured and described?
2. Methodology The approach is simple. When you measure utilisation of carbon source, production of biomass and other products and growth rate then, provided you also know the metabolic routes which operate, you
have all the data required to construct a quantitative description of all the reactions (net) which sustain these processes. Such a description contains the flux through every enzyme in the CMPs and is called a flux analysis. 2.1. Definitions and units The basic parameter of a fermentation is the biomass, expressed as dry weight-usually kilogram dry weight biomass (kg dry wt). In the end, everything else is related to this. All other measurements are made in moles which avoids the confusion of changes in weight when substrates differ from product by virtue of phosphorylation, oxidation, decarboxylation, etc. An 'input' is the amount of a feedstock which is taken up and metabolised by the biomass. Thus, when E. coli is growing on glucose as the sole carbon source, we measure the input as the amount of glucose required to generate 1 kg dry wt. Throughputs are also amounts (in the same units) which are defined as the amount of a substrate (e.g. fructose 1,6-bisphosphate) converted to products (the triose phosphates: C3P) in generating 1 kg dry wt. Precursors are converted to monomers contained in the end-product (biomass) and both are in the same unit. There are, of course, other products such as CO 2 and sometimes other excreted products which also are expressed in the same units. In the steady state, the biomass generates itself at a constant rate described by the growth rate /x (per hour). The product of the amounts defined above and /z are m o l / k g dry w t / h and these are fluxes which describe the rate at which an input is used, the flux by which a substrate is converted to product or a monomeric constituent of biomass is generated or another product such as acetate is excreted. The last unit of possible significance in flux analysis is pool size which is an amount expressed as moles of an intermediate contained as a pool within the biomass and again has the unit m o l / k g dry wt in any given steady state. The rate at which the intermediate must be generated to maintain pool size is derived from /x in the same way to give moles pool intermediate/kg dry w t / h . In prokaryotes where pool sizes are very variable and usually small relative to their throughput, this value has limited importance and the turnover of the pool is probably more significant.
89
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-1 / 6
This is the rate of input or output (which are the same) divided by the pool size, i.e. m o l / k g , dry w t / h / m o l / k g • dry wt which has the dimension of a number/h and is the number of times the pool turns over every hour. It is obvious that in a linear segment of metabolism where the flux through each enzyme in the chain is the same, then the turnover of each pool of intermediates is inversely proportional to the size of the pool (this is discussed in an example below, Section 3.3.6.).
Glucose6P ,
M°D~ii! cetone:r u~'6~ilcera~ldehyde3P
/
DiPGlycerate
2.2. From inputs to precursors
3PGlycerate~ >Monomers
E. coli ML308 converts a single carbon source to biomass, carbon dioxide and, sometimes, acetate.
Glucose
h~Monomers
~ Glucose6Pi ~Monomers
2PGlylerate Monomers *~ Monorners~
~PenolPyruvate
Py ~PC /Malate ~4 ~
,,JPyruvate
Oxalac( :ate
Fumarate
h2~lonomers
F;iiioiil;; z i' DihyxyacetoneP -~-~lyceraldehyde3P Monomers
DiPGlycerate 3PGlycerate~ ~,Monomers 2PGlylerate
Monomers< Monomers <~
Pyruvate
~Penol
,Pyruvate
OxalaeE tate
Fumarate
M....... <~D~coA 'J~C i~1ate~M°..... Succ!nate iso
itrate Sy!ICoA '~ Glutarate, Oxo ~ Monomers
Fig. 2. Pathwaysto precursorsof monomersduringgrowthon
glucose. PDH: pyruvate dehydrogenase;PEPC: phosphoenol pyruvatecarboxylase;PK: pyruvatekinase;PPP: pentosephosphatepathway;PTS:phosphotransferasesystem.
M....... <~D~coA ~Ci] 'ate ~M . . . . . . . Succmate ; iso
Citrate Su~ylCoA ~P~Glutarate Oxo [ =~Monomers
Fig. 3. Pathwaysto precursorsof monomersduringgrowthon glycerol. The CMPs receive the carbon input and direct the flux to all the outputs. The large number of monomers from which biomass is assembled are made by known biosynthetic routes served from a rather limited number of precursors. These are usually the same, irrespective of the carbon source because the biosynthetic route to any given monomer usually starts from the same precursor. However, there is one exception in that the routes to C 4 and C 5 sugars are taken from the PPP which can be fed either by oxidative decarboxylation of 6-phosphogluconate or by rearrangement of C3P. In addition, the fluxes through the CMPs are modified to accommodate provision of precursors from different inputs, for example, gluconeogenesis from pyruvate requires a unique enzyme to generate phosphoenol pyruvate (PEP) and then reversal of glycolysis (using fructose
90
H. Holms
/ FEMS Microbiology Reviews 19 (1996) 85-116
bisphosphatase rather than phosphofructokinase) to generate glucose 6-phosphate (G6P). However, as far as carbon fluxes are concerned, a principal function of the CMPs is to sustain fluxes to the precursors for biosynthesis and these must be considered relative to the carbon input. The first task for flux analysis is to construct a diagram of the CMPs for each carbon input under study. Such diagrams are shown for glucose (Fig. 2), glycerol (Fig. 3), pyruvate (Fig. 4) and acetate (Fig. 5). The pathways from precursors to monomers are found in the standard texts and, more particularly, in an earlier systematic treatment of the basic principles of flux analysis [4]. The precursors used [4] to make biomass are G6P, pentose phosphates (C5P), tetrose phosphate (C4P), glyceraldehyde, 3-phosphate (C3P), phosphoglycerate (PG), PEP, Pyr, Ac.CoA, oxaloacetate (OAA)
Glucose6P~
~
s
' ~,M. . . . . . .
Monomers
/
,
Monomers
PenolPyruvate~ S~nE~if~ E P ( /Malate ~
x
Pyruvate
//PO.
Oxalac ate
Monon,ers*
AeetylCoA
.~Monomers
2PGly!erate
Monomers
<~
PenolPyruvate
Pyruvate
\
/MO.(dc)
/ Ae?tyIP
/
Citrate :1" Glyoxylate A ~ ICLY
isoC~ra te~
_Succunate
Succi!ylCoA
~P~GluOtX°ate/,
M. . . . . . .
.~Monomers
2PGly(.=rate
~P~
/
DiPGlycerate
Fig. 5. Pathways to precursors of monomersduring growth on acetate. ICL: isocitrate lyase; MDH (dc): malate dehydrogenase (decarboxylating); ME: malate enzyme; PEPCK: phosphoenolpyruvatecarboxykinase.
DiPGI' ,eerate
External Pyruvate
~ -D~Glyeeraldehyde3P
3PGlycerate [
Acetate
rU ~ G I I I I c ~ hyde3P
3PGly :erate
roxyacetoneP
. ....... :'=%cet,
F t t 16d,P ~
Monomers
~Monomers
Fruct( ~l.6diP/'
Fruct!se6P%~hf~
DihydroxyacetoneP
se6P%~l
Fruc'
~ ~Monomers
Glucose6P
~ Monomers
Fumarate
M...... s
Suc( ~ate
Cit 'ate
Monomers iso Citrate
Succi
ICoA
\oxo/r ~ Glutarate
Monomers
Fig. 4. Pathways to precursors of monomersduring growth on pyruvate. PEPsyn:phosphoenolpyruvatesynthetase.
and OGA. All of these precursors must always be provided to generate new biomass and our four models achieve this by minor modification of the CMPs and unique mechanisms to take up the particular carbon source and deliver it into the CMPs. These can be briefly described (for pathways see [1-4]). Glucose is taken up and phosphorylated by the phosphotransferase system (PTS). Glycerol (Fig. 3) is taken up by facilitated diffusion, gives G6P by gluconeogenesis but can feed the PPP from C3P without oxidative decarboxylation. Pyruvate (Fig. 4) is taken up by a proton symport and requires PEP synthase to generate PEP. Acetate (Fig. 5) is taken up by a proton symport, phosphorylated by acetokinase to acetyl phosphate and thence to Ac.CoA by a phosphotransacetylase. All other precursors, in the acetate
H. Holms / FEMS Microbiology Ret,iews 19 (1996) 85-116
pbenotype, depend on the glyoxylate bypass which generates malate and thence OAA from Ac.CoA by ICL and malate synthase (Fig. 5). How do we measure the fluxes from precursors to biomass? OGA is the precursor for four amino acids: glutamate, glutamine, proline and arginine. The content [4] of these amino acids in glucose-grown E. coli ML308 is (tool kg - ~ dry wt) Glu 0.353, Gln 0.201, Pro 0.252 and Arg 0.252 and as each derives from 1 tool OGA it follows that an output of 1.058 tool OGA is required to make 1 kg dry wt biomass. At a /z of 1.0 the flux of OGA to biosynthesis is also 1.058 mol kg-~ h -~. In the same way, the output of all the other precursors is calculated from the monomeric composition of the biomass and the growth rate (Table 1) on any different carbon sources from measured data [4,13,14].
91
" Gluelose6P~.'.'.'.'.'.'.'.'.~M.......
TS•
Glucose
Fructose6P
Fru~bisP TriosePs~
Monomers
¢-7
~
Monomers
yruvate~
Monomers
3PGlycerate
Peno
Monomers~
Pyruvate
Oxal ~
Flux analysis is simply a method of organising data and these are taken, in the first instance, from Table 1 and applied to the metabolic maps defined in Fig. 2, Fig. 3, Fig. 4 and Fig. 5 for flux of carbon only. All cofactors are taken from pools which, in the steady state, are maintained constant by regeneration and. as there is no net change, are ignored. Furthermore, as the flux through each step of a linear segment is the same, many of these are omitted and only where fluxes are changed by removal of precursors or excreted products or at branch points do they need to be included. The conversion of data to a flux
Monomers
Acetate
2.3, Flux analysis of growth on glucose
M....... ~ r ~ . ~ _ A~:~
__L__ C.!ate ,
,
AOP~ArPAcetaisoteCitrao~o te
o,o,.rt&
Excrete
Monomers
Fig, 6. Flux diagram for growth of E. coli ML308 on glucose showing only those fluxes directly derived fiom measured data (mol/kg dry wt/h).
Table I Fluxes (mol/kg dry w t / h ) from precursors to monomers from six carbon inputs at various growth rates (/x) Precursor Glucose 6(P) Triose(P) (P)Glycerate (P)enol pyruvate Pyruvate Oxalacetate Oxoglutarate Acetyl CoA
Carbon source (#)
(I,00)
Glucose (0.94)
Gluconate (0.90)
Fructose (0.72)
Glycerol (0,70)
Fumarate (0.63)
Acetate (0.43)
1.026 (1.984) * 0.140(1.636) + 1,362 0.554 2.318 1.688 1.058 2.643
1.86 ~ 0.13 1.28 0.52 2.18 1.59 0.99 2.48
0.92 ~ 0.13 1.23 0.50 2.09 1.52 0.95 2.38
0.74 1.18 + I).98 0,40 1.67 1.22 0.76 1.90
0.72 1.15 + 0.95 0.39 1.62 1.18 0.74 1.85
0.65 1.03 + 0.86 0.35 1.46 1.06 0.67 1.67
0.44 0.70 (}.59 0.24 1.00 0.73 0.45 1.14
Flux to PPP from G6P. + Flux to PPPs from triose(P) (see Fig. 18),
92
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
TS•
diagram requires three stages. Firstly, a diagram specific to the carbon input and the routes to precursors must be constructed. Secondly, the measured data are expressed as fluxes for input and provision of precursors for monomer biosynthesis and entered into this diagram. The simplest basic flux diagram for growth on glucose incorporates these two stages and includes input, output of precursors, acetate excretion and the inevitable consequences of these measured data-the flux of PEP to pyruvate for glucose uptake by the PTS and the anaplerotic flux to the Krebs cycle for provision of OAA and OGA (Fig. 6). This figure illustrates the principles of flux analysis which can be summarised as follows: when you measure everything that goes into the CMPs and everything that comes out, then you can construct a flux diagram provided you know the metabolic routes used. The fluxes are written in boxes as mol/kg dry wt./h. The flux diagram is then completed (Fig. 7) by simple arithmetic which is very straight forward and will be illustrated by only a few salient points. The flux of glucose into the cell is 10.57 of which 1.86 fluxes to monomers leaving 8.71 for glycolysis. Aldolase splits FDP to 2 mol of C3P of which 0.13 is used as precursor for lipid biosynthesis. A further 1.28 is removed to monomers as PG leaving 16.01 as the flux to PEP. Of this amount 10.57 is fluxed to pyruvate to donate phosphate for glucose uptake, 0.52 to biosynthesis of aromatic amino acids, 2.58 for anaplerotic provision to the Krebs cycle leaving
= Glucose6P "~'-~l~-~Monomers I
Glucose
Fructose6P
Fru~bisP TriosePs ~
Monomers
3PGlycerate ~
Monomers
PenolF~yru~at~e M. . . . . . . ~ l ' ~ ' 8 - ~ P y r u y ate
M. . . . . . . " ~ - ~ ' - -
~""'~"
Monomers
AcetateOXal~
t t r~j
M. . . . . . .
~
AOPATPAcet ~ ateisoCitrao~o te Glutarate Excrete
Fig. 7. Complete flux analysis of growth of E. coli ML308 on glucose. Fluxes are mol/kg dry wt/h.
2Oxoglutarate ,coo.
Glutamate coo.
C=Oi
CHNH2
COOH NA
+
COOH AcCoA
Glutamine
,coo.
coo.
N-AcetylgJutamyl semlaldehyde c,~) coo.
CHNH-COCH 3
CHNH-COCH 3
j/~ ~ CHNH_COCH 3
N*Acetylglutamate N-Acetylglutamyl(P)
CoA
COOH ATP
ADP COO(P) NADPH,H + NADP+ CHO
Proline OGA
COOH
F. . . . . .
COOH COOH
~..,
.~H,
CH2NH-- ICI--NH2 NH Arglnlne
CH2NH--C:NH
~ A s p
.NH
Argininosuccinate
COOH (~) ~.,
AMP (P'P)
ATP CH2NH.CONH 2 Citrulllne
COO/P ) COOH Acetate COOH CHNH2 L CHNH-COCH, ~,H2 < ¢,H, CH2NH2 Ornithine
Fig. 8. Synthesis of the glutamate family of amino acids.
CH2NH2 N-Acetylornithine
H. Holms / FEMS Microbiology ReL'iews 19 (1996) 85-116
Carbonsource OXOgi#/u~'u,rPar,:c ur. . . . C
~
D
P
.
Arginine
6ATP AMP 4NADPH,H+ "Energy" 5(P) NAD+ (P.P) 4NADPH,H ÷ ~ A D H H ,
Carbon source ÷ 02
÷
~ Carbon dioxide + H20
Fig. 9. Net reactions in CMPs during synthesis of arginine from oxoglutarate.
2.34 to be converted to pyruvate by pyru~zate kinase (PK). The flux to pyruvate is 12.91 (10.57 + 2.34) of which 2.18 is used for biosynthesis. The remainder (10.73) is decarboxylated by PDH to Ac.CoA of which 2.48 is used for biosynthesis, 4.89 fluxed to acetate excretion and the remainder (3.36) fluxed into the Krebs cycle by citrate synthase which also uses 3.36 m o l / k g dry w t / h of OAA made up of the fluxes to OAA (2.58 by PEPC and 2.37 by operation of the cycle) less 1.59 removed for biosynthesis (2.58 + 2.37 - 1.59 = 3.36). This diagram (Fig. 7) describes all the fluxes through the CMPs when E. co/i ML308 is growing on glucose. Because we know the biosynthetic routes from precursors to monomers it is possible to extend the flux analysis to each monomer. This would generate a very large flux diagram and only one will be used as an example! The flux of OGA to the glutamate family is 0.99 to Glu of which 0.332 goes to protein biosynthesis, 0.237 to Pro, 0.188 to Gin and 0.237 to Arg. Again the fluxes to all of these can be calculated, but Fig. 8 gives only Arg in detail. The overall equation for the biosynthesis of 1 mol Arg is:
93
taken from the metabolic pools within the cell. These metabolic pools are kept constant by regeneration. Carbamoyl (P) is made from CO 2 + 2ATP by a synthetase using Gin as amino donor which is, in turn, regenerated from Glu, NH 3 and ATP by Gin synthetase: CO 2 + 3xATP + NH 3 = Carbamoyl(P) + 3xADP + 2x(P) Acetyl CoA is regenerated from acetate, ATP and CoA by acetokinase and a specific transacetylase: Acetate + ATP + CoA = AcCoA + ADP + (P) Glu is regenerated from OGA, NADPH,H ÷ by Glu dehydrogenase:
NH 3 and
OGA + NH 3 + NADPH,H ÷ = Glu + NADP ÷ Asp is regenerated from fumarate, H20, NH 3,
Giucose6P ~
Glycerol
Monomers
Fructose6P
GlyceroP
\/
- TriosePs~
Monomers
3PGlycerate ~
Monomers
yruvae ~ "P,~t
Monomer8
Penol
M....... ~ P y r u v a t e
AcOetaale~
M. . . . . . .
OGA + NH 3 + 2xNADPH,H+ + 2xATP + AcCoA + Glu + Carbamoyl(P) + Asp = Arg + 2xNADP ÷ + ADP + AMP + 2x(P) + (P.P) + Acetate + Fumarate The inputs are OGA and CO 2, both of which are generated by the CMPs, ammonia which is a component of the medium, together with NADPH,H ÷, ATP, AcCoA, Glu, Carbamoyl (P) and Asp which are all
AcetyI.P
[~
ADP~ATPAceiso tateCitraox~o te
~ Glutarat~_~
Excrete
Monomers
Fig. 10. Flux analysis of growth of E.coli ML308 on glycerol.
94
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
NAD ÷ and NADPH,H ÷ by fumarase and malate dehydrogenase to give oxalacetate which is transaminated to Asp by Glu which, in turn, is regenerated by glutamate dehydrogenase:
tion of the CMPs: provision of precursors and the energy required to convert them into monomers and thence biomass.
Fumarate + H 2 0 + NH 3 + N A D + + NADPH,H +
2.4. Flux analysis o f growth on glycerol
= Asp + NADH,H+ + NADP + Overall the conversion of the precursor, OGA, to the monomer, Arg is according to this equation: OGA + CO 2 + 4xNH 3 + 4xNADPH,H ÷ + NAD + 6ATP = Arg + 4xNADP÷ + NADH,H ÷ + 5xADP + AMP + 5(P) + (P.P) The overall balance of the various reactions is gathered in Fig. 9 to show the dual function of the CMPs. Flux through the CMPs generates the two precursors, OGA and CO 2, required for Arg biosynthesis. In addition the CMPs provide the 'energy', reducing power and ATP, required to drive Arg biosynthesis, a large part of which is used to regenerate the pools of intermediates also fed into the process. Finally, in the catabolic role, the CMPs must regenerate the pools of ATP, NAD + and NADPH,H +. This illustrates very well the dual func-
CHO ~Ii=O CH)O(P)
.~
Glyceraldehyde3(P)
CH2OH C=Oi HO--C-H i H-C-OH i H-C-OHI CH20(P)
Growth on glycerol is slower (74%) than on glucose, and fluxes [4] reflect this (Fig. 10). Furthermore there is no acetate excretion and the sole products of growth on glycerol are biomass, CO 2 and water. This is discussed more fully later (Section 3.3.3.). Although the input of glycerol is greater on a molar basis, the delivery of carbon into the CMPs in only 66% of that on glucose. The uptake of glycerol is by facilitated diffusion and one specific enzyme (glycerokinase) followed by glycerophosphate dehydrogenase to yield C3P. From this junction, gluconeogenesis is by fructose bisphosphatase rather than the phosphofructokinase used on glucose. Otherwise routes to precursors are the same as for glucose except, of course, there is no requirement of PEP for uptake and the flux into the PPP to generate C5P and C4P is presumed to be from C3P (Fig. 11). Proportionally less CO 2 is generated by PDH and more is generated in the Krebs cycle.
CH2OH C=O -dr Trbnsketolase ~
~ ~
Fructose6(P)
-~
CHzO(P)
Glyceraldehyde3(P)
ICH20H C=O I HO--C-H I
H-C-OH
CH20(p)
Xylulose5(P)
~
Transaldolase
~
CH20(P)
"I"
Rlbose5(P)
"4"
CH20(P)
Sedoheptulose7(P)
CHO H-C-OH I H-C-OH H-C-OH ~ ' ~ CH20(p )
H - C,- O H H--C-OHm
~ F
Erythrose4(P)
CH2OH C=O HO--C-H I H-C-OH I H-C-OHm
Clio
C=O
I CH20(P )
CHOI H--C-OH , H--C-OH CH20(P)
Fructose6(P)
Transketolase
CHO 1 H-C-OH i H-C-OHI CH20(P) Efithrose4(P)
CH2OH C=O HO--~:-H t
I
CHO C=O,
~-
Xylulose5(P)
CH2OH C=O HO--C-H H - Ci- O H H-C-OH H-C-OH
HO--C:-H H - Ci- O H CH20(p )
"I"
Glyceraldehyde3(P)
H-C-OH CH20(P)
Sedoheptulose7(P)
Fig. 11. Nonoxidativeconversionof glycolyticintermediatesto C4 and C5 sugars in the PPP.
H. Holms / FEMS Microbiology Rm,iews 19 (1996) 85-116
Glucose6P"~O"~5-~ Monomers Fructose6P
TriosePs~
Monomers
3~rate ~
J
M. . . . . . . ~
excreted as acetate and 38% fluxed to citrate in the Krebs cycle and thence 0.9% into OGA (the flux to OAA is via PEP). Compared with glucose, a large amount of pyruvate is oxidised by PDH and the Krebs cycle to CO~ and a very large amount of carbon is excreted as acetate. 2.6. Flux analysis of growth on acetate
I
Pyruvate
95
Monomern
Peno~vruvate~ PepSynJ
~
~ru~ate
Monomers PEPC
.=l~te ~
M....... ~ ~ A 3 ( ~e~t~[~
M. . . . . . .
Growth on acetate is slow (/~ = 0.43) and delivery into the CMPs is by acetokinase and a phosphotransacetylase to generate Ac.CoA. This is itself a precursor for biosynthesis but all other precursors must be generated from it by flux through the glyoxylate bypass (Fig. 5). This route, unique to growth
~
T;;/
Glucose6P"~-~4--~- Monomers Fructose6P
/
I
AOPl
iso
Citrate
FructosebisP
Acetate Oxo
°'u'srate--
TriosePs~
Exc~;ete Excrete u.ttt_2zLL_~Monomers Fig. 12. Flux analysis of growth of E. coli ML308 on pyruvate.
Growth on pyruvate is as fast ( / x = 0.73) as glycerol but flux analysis [13] shows (Fig. 12) big differences between pyruvate and glycerol and glucose. The uptake of pyruvate is very large and, on a carbon basis, is 3.t times greater than glycerol and twice that for glucose. 14.2% of the pyruvate uptake is phosphorylated to PEP (by PEP synthetase) and used directly for biosynthesis, anaplerotic provision to the Krebs cycle or gluconeogenesis by reversal of glycolysis for generation of precursors for biosynthesis. Of the pyruvate taken up (100%) a small amount (I .4%) is reduced to lactate and excreted and 3.9% is used for biosynthesis. The rest (81.9%) is oxidised to Ac.CoA. Relative to the original pyruvate uptake, 4,5% of the Ac.CoA is used for biosynthesis, 39.2%
3~rate ~
Monomers
~
Monomers
PenolPyruvate
2.5. Flux analysis o[ growth on pyru•ate
M....... ~
Pyruvate~. ~
Monomers ~
Acetate
[ ~
M.......
~Oxal
A~ety[
/~
Monomers
Acetate
ICx~O~'Acetate/s°lCitr~~' e~Glutarate *~umarate~e ~
Monomers
Fig. 13. Flux analysis of growth of E. coli ML308 on acetate.
96
H. Holms / FEMS Microbiology Reuiews 19 (1996) 85-116
Glu~ose6P~
on acetate and other 2-carbon inputs, generates malate and thence O A A which flux to Pyr and PEP, respectively. On these assumptions the measured data can be calculated as a flux analysis (Fig. 13). The implications of these assumptions must be clearly defined. These are: • the flux to A c . C o A is directly from the carbon
"l'riosePa~
3~rate
Pyruvate
~
Morlorner8
Monomers
~
Monomers
PenolPyruvate---~.~-~M . . . . . . .
Oxal
P ruvate
Acetyl~
8.471
~' Glucose6P-~'-~3"~Monomers
TS•
L
Fructose6P
Glucose
uctob~se ~--]-~-----
~t = 0 . 7 2 Chemostat
Fr
L ctate
I
/ ~ ......
Monomers
j Peno~ate ~
Monomers
~Pyru~ate
AcOe~ate~
Acetate
Excrete
Fig. 15. Flux analysis of E. c.oli ML308 in a pyruvate chemostat at /x = 0.30.
M. . . . . . .
~
ADP p~ AT
Acetate 0~o - ~ Glutarate
,
3PGlycerate ~
~
~,._
AcetyI.P ADP= iSOCitrate
isP
Excrete
M....... ~ 0 - " ~
.......
Fructoae6P
source;
• there is no need for a flux from pyruvate to A c . C o A and PDH does not operate; and • the route to pyruvate is from malate by malate enzyme (ME) and to PEP is from O A A by PEP carboxykinase (PEPCK). This illustrates a fundamental problem in the technique of flux analysis. W e need to know the fluxes from precursors to monomers and the routes by which the carbon source sustains these fluxes. The
. .......
M
isoCitraox~o te Glutarate
Excrete
Fig. 14. Flux analysis of E. coli ML308 in a glucose chemostat at /z = 0.72.
flux from precursors is no problem because it is derived from a measurement of the monomers generated for biosynthesis. In this case, however, the metabolic routes employed are a matter of judgement. W e know, for certain, that there must be a flux of 1.00 to pyruvate and o f 2.41 to PEP and we choose to make these by ME and P E P C K respectively. However, in principle, at one extreme there could be a flux from O A A to PEP of 3.41 and thence, to pyruvate the flux o f 1.00 by pyruvate kinase. At the other extreme, there could be a flux of 3.41 from malate to pyruvate and thence to PEP a flux o f 2.41 by PEP synthase. In between, there could be any combination of these fluxes. W e choose the simplest, most economical solution which is an anthropocentric view of what E. coli would be best advised to adopt! However this may be, there is no
97
H. Holms / FEMS Microbiology ReL,iews 19 (1996) 85-116
doubt that the net fluxes to pyruvate and PEP must be 1.00 and 2.41! The unique features of flux analysis of growth on acetate are obvious: • flux to all precursors, except Ac.CoA, is by the glyoxylate bypass which must compete with flux through the Krebs cycle. In particular, ICL and ICDH compete for isocitrate; and all fluxes to precursors differ from those operating on glucose and indeed many other carbon sources.
Fructose
PTS
Glucose6P
+
'
[ ~
Fructose6P
FructoselP
Fructose bisP
TrlosePs~
2.7. Flux analysis of growth in chemostats and on other carbon sources
Flux analysis of growth on the other carbon sources and conditions will be presented in less
"~-0"~-]--'~" Monomers
Monomers~
Monomers
3PGlycerate ~
Monomers
Peno/P~yruvat~e
Monomers
- Pvru~a' yru]t e
Oxal Acetate
~
Monomers
= Glucose6P~ M o n o m e r s
Prot
on•sym port
Glucose 6P
I
[ ~
Fructose6P iso Citrate
Fructose bisP
ATP
Acetate
TriosePs~
Monomers Excrete
.......
3PGlyeerate~
Monomers
Pone yruvate~
Monomers
]
Acetate ~
Monomers
Fig. 17. Flux analysis of growth of E. coli ML308 on fructose.
Monomers
.......
AOPATPAcet ~ ateisoCitraox~o te Glutarate Excrete
Fig. 16. Flux analysis of growth of E. coli ML308 on glucose 6-phosphate.
detail. When uptake of glucose is restricted in a chemostat, acetate excretion is diminished and, at /x of 0.72, is zero (Fig. 14). Under these circumstances there is no flux through pyruvate kinase and the flux to PEP is used for uptake (73.2%) biosynthesis of aromatic amino acids (4.7%) and anaplerotic flux into the Krebs cycle (23.3%) which add up to rather more than 100% because the data have been 'rounded up'. The flux from glucose to PEP is insufficient and the remainder is shown to be generated by recycling pyruvate through PEP synthetase. This is necessary to balance the flux diagram but, of course, data are not sufficiently accurate to predict such a flux. It is more probable that there is no reverse flux through PEP synthetase which normally does not operate in a glucose culture. Nevertheless this illustrates a problem inherent in this form of flux analysis, or, indeed,
98
H. Holms / FEMS Microbiology Ret,iews 19 (1996)85 116
in any other method such as metabolic control analysis, both of which assume that we know the detail of the metabolic routes actually used in the system under study. Measurements of inputs and outputs derive net fluxes of substrate to product. If PEP synthetase was expressed in the glucose phenotype and if it did operate, a futile cycle would be possible and the flux analysis would not detect it. In the same way when pyruvate uptake is restricted in a chemostat, acetate excretion is diminished and at /z = 0.30 is totally abolished (Fig. 15). While growth rate falls to 41% pyruvate uptake is only 25% of that when unrestricted. Flux of precursors to monomers follow the fall in growth rate but oxidation of Ac.CoA in the Krebs cycle is actually increased, relative to /~, by 26% in the chemostat as compared to the batch. This increase in oxidative
Glucose6P ~
Glucuronate
Monomers
[ ~ Fructose6P
Mannonate :
\/
P\
PenolP~rLvate~
Monomers
.......
.......
Glucose6P ~-[-0"~'9"2-'~ Monomers Gluconate Acetyl.P
1
Aop~ArPAcetate
Fructose6P
isoCitr@e
6P-Gluconate
/"
~
\/
Monomers ~ 2 Keto 3 deoxy
TriosePs ~ /-~¢~_._~
Monomers
Excrete
Glutarat[~,.
Monomers
Fig. 19. Flux analysis of growth of E. coli ML308 on glucuronatc.
\
PenolPyruvate ~
Monomers~ '~''~-~ ' p- Y•ruvat Ve I
Monomers
Oxal ~ u Acetate f .... i ~ monomers
AOPATPAcet ~ ate iso Citr@te Glutarate
Excrete Fig. 18. Flux analysis of growth of E. coli ML308 on gluconate.
energy production presumably replaces the ATP generated by acetate excretion in the batch. In other words, restriction of pyruvate uptake permits it to be partitioned between the two functions of the CMPs (precursors and 'energy') in the most economical fashion. Glucose 6-phosphate is taken up by a proton symport but otherwise pathways are the same as for glucose (Fig. 16). Growth rate also is similar but 11% more carbon is taken up, t8% less acetate is excreted but 1.9 times as much Ac.CoA is oxidised in the Krebs cycle. Fructose is taken up by a PTS to generate fructose l-phosphate and, on the principle that the shortest and most economical routes to precursors are chosen, triose phosphate is shown as the route to C 5 and C 4 sugar phosphates by the PPP (Fig. 17). No acetate is
99
H. Holms / FEMS Microbiology RetJiews 19 (1996) 85 116
excreted and growth rate is slower than on glucose but equal to that in the glucose chemostat at which acetate excretion is also zero. To all intents and purposes, a fructose phenotype at /'/'max is metabolically equivalent to a glucose chemostat at /z0.72. However, the carbon uptake is 1.23 times greater and oxidation of Ac.CoA in the Krebs cycle 2.22 times greater than the glucose chemostat. Obviously the conversion of glucose to biomass at /~0.72 is much more efficient than fructose at the same growth rate. Flux analyses show that the glucose chemostat (Fig. 14) generates 4.69 tool A T P / k g / b (net) by glycolysis while the fructose culture (Fig. 17) makes 14.18 tool A T P / k g / h (net) as well as proportionally more reducing power. There is therefore no obvious reason that the fructose phenotype has to oxidise so much
Glucose6P- ~ - ~ - ~
M. . . . . . .
Fructose6P
TriosePs
~
Monomers
3~rate ~
Monomers
PenoIPy~vate~ Monomers~
Pyruvate [~ ~
Glucose6P----[O~]--~- Monomer s
Monomers
~
Oxal - Acetate~ 0.54~'~ M.......
Citi'ate
Malate
AcetyI,P [ ~ isoCitrate
Fructose6P
Fumarate
ATP
Acet 4 at Lactate
TriosePs ~
Monomers
Penoyruvate~
Acetate
~f Acetate
Monomers
isoCitrate Ox~o
Excrete
Fig. 20. Flux analysis of growth of
Fig. 21. Flux analysis of growth of rate.
E. coil
ML308 on oxogluta-
Monomers
Oxal ~
Monomers
Excrete
Monomers
+ 3~rate ~
LDH
Pyruvate
Gl~~r~sUccin2oglutarate
\/
Monomers
E. co/i
ML308 on lactate.
more Ac.CoA in the Krebs cycle. This problem is discussed further below (Section 3.3.4.). Gluconate (Fig. 18) is taken up and phosphorylated to 6-phosphogluconate which is a component of the PPP and presumed to be the source of pentoses and tetroses for biosynthesis. The EDP then generates equal amounts of triose phosphate and pyruvate. The consequence of this is that only 32% of the gluconate-carbon reaches PEP but a large fraction is dephosphorylated by PK and joins the very large flux from EDP to pyruvate. The consequence is a very large acetate excretion. Glucuronate (Fig. 19) is also metabolised by EDP and, although growth is slower (/x = 0.87) than on gluconate (/z = 0.90), acetate flux to excretion is 40% greater, and oxidation in the Krebs cycle is 31%
100
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
less. On the other hand, metabolism of glucuronate to 2-keto-3-deoxygluconate generates reducing power but, because the route does not pass through 6 phosphogluconate, precursors from the PPP are generated by the non-oxidative route. Lactate (Fig. 20) is obviously similar to pyruvate (Fig. 12) but uptake and growth are slower. Acetate excretion much less (17% of uptake, on a molar basis, compared with 39% for pyruvate). Here again is a pair of carbon sources with almost identical metabolic profiles. Lactate compared with pyruvate is slower in growth (82%), acetate excretion (24%) and oxidation of Ac.CoA (71%). The explanation must rest on the slower uptake (55%) of carbon source and the additional reducing power generated by oxidation of lactate to pyruvate. Growth on OGA (Fig. 21) is very slow (/z = 0.32) and no acetate is excreted. Uptake of carbon source is very slow and probably limiting (see Section Glucose6P
""~'-~'.6-5-~--~-Monomers
Fructose6P
TriosePs ~
Monomer=~
Monomers
3~rate
~
Monomers
Penol Py[~ate
~
Monomer,
Pyruvate
l
Acetate
Excrete
'~
"
~
Oxal - A~e;;te --~ 1.06 ~
/C.ta,o
M. . . . . . .
.um r=o
.t~o~$uc¢inate Glutarate ~
Monomers
Fig. 22. Flux analysis of growth of E. coil ML308 on fumarate.
3.3.2.). Oxoglutarate is the only carbon source in the set of eleven on which E. coli ML308 will not grow without a genotypic change (constitutive expression of the dicarboxylic acid permease). Fumarate (Fig. 22) on the other hand, supports a reasonable growth rate (/x = 0.63) without acetate excretion.
3. Discussion 3.1. The distribution of carbon fluxes in the CMPs The fluxes of the carbon sources through the CMPs (Table 2) show the extraordinary flexibility of the system. Where thermodynamics and pool sizes make a reaction physiologically irreversible, there is usually another enzyme which sacrifices 'energy' in order to drive the reaction in the 'opposite' direction. In this sense, probably only PDH and OGA dehydrogenase are irreversible. There is a large gap between the phosphorylated and non-phosphorylated parts of the CMPs. The pool of PEP is at the centre of this junction and the fluxes in and out cover a very wide range depending on the carbon source. Enolase, PEP synthetase and PEPCK generate PEP within the range of 7.36-18.3, 0-6.1 and 0-3.5 respectively. Fluxes out are by 'reversal' of enolase to feed gluconeogenesis (1.5-6.1), PTS for uptake (0-10.6), PK (0-15.2) and PEPC for anaplerotic provision to the Krebs cycle (0-2.6).Whatever the nature of the carbon source or special circumstances of any one experiment, the function of the CMPs is always to sustain fluxes to precursors and the energy required for biosynthesis. The fluxes to precursors are a question of measurement as are the other carbon outputscarbon dioxide and, on some carbon sources, acetate. The energy required for conversion of precursors to monomers and their assembly into biomass is generated by the CMPs as a small amount of ATP and a much larger amount of reduced nucleotides ([2H]). Some [2H] is used to make monomers more reduced than their precursors and, as we know the biosynthetic pathways this is subject to calculation. We do not know the amount of ATP made by oxidation of the balance of [2H] available but it must be at least sufficient to power biosynthesis and growth. This will be dealt with later (Section 3.3.4.). The data from the flux analyses (Fig. 7, Fig. 10,
101
H. Holms / FEMS Microbiology Reviews 19 (1996) 85- 116 Table 2 Fluxes through key enzymes with CMPs on eleven carbon sources
G6P Glucose Gluconate Glucuronate Fructose Glucose chemo Glycerol Pyruvate Fumarate Lactate Acetate Pyr chemo Oxoglutarate
Growth rate /*
PTS
PFK
FbisP ase
TPDH
Enolase
PK
PEP Syn
PDH
PEPC
PEPCK
ME
Acetate excretion
(I.95 0.94 0.90 I).87 I).72 11.72 0.70 11.73 I).63 0.60 0.43 0.35 0.32
0 10.57 11 1) 7.62 6.22 0 0 0 1) 11 11 0
9.88 8.71 0 0 0 4.79 0 11 11 0 0 0 11
0 0 11.92 0.89 0.74 11 0.72 0.75 11.65 0.62 0.44 I).36 0.34
19.63 17.29 8.65 8.54 12.58 9.48 11.41 - 2.69 - 2.33 - 2.22 - 1.58 - 1.29 - 1.20
18.34 16.01 7.42 7.36 I 1.6 8.5 10.46 - 6.08 - 3.19 - 3.04 2.17 - 1.53 - 1.64
t5.211 2.34 4.45 4.49 1.60 0 8.15 0 0 0 0 0 0
1l 1) 0 0 0 (I.I 0 6.118 0 5.01 0 2,92 1/
13.00 10.73 12.98 14.21 7.55 4.45 6.53 35.111 9.79 17.26 0 10.02 0
2.61 2.58 2.47 2.39 1.98 1.98 1.92 2.(X) 11 1.64 0 I).96 0
0 (I 0 0 0 11 11 (1 3.54 0 2.41 0 1.82
I) 0 0 11 0 0 0 I) I 1.25 0 1.00 0 3.95
4.(10 4.89 6.48 9.117 II I) 0 16.81 0 4.06 -20.16 11 0
FbisPase: fructosbisphosphatasc. TPDH: triosephosphate dehydrogenase.
Fig. 12, Fig. 13, Fig. 14, Fig. 15, Fig. 16, F i g . 17,
a l s o m a d e by o x i d a t i o n o f i n t e r m e d i a t e s to t h e car-
Fig. 18, Fig. 19, Fig. 20, Fig. 21 a n d F i g . 2 2 ) are
b o x y l i c a c i d s w h i c h are t h e s u b s t r a t e s f o r d e c a r b o x y -
a s s e m b l e d in T a b l e 3 b y g r o w t h rate. F l u x o f s o m e
lation. In a s e n s e , f l u x to c a r b o n d i o x i d e r e f l e c t s t h e
c a r b o n s o u r c e s into t h e C M P s e x c e e d s t h e f l u x e s to
p o t e n t i a l o f t h e C M P s to t r a p ' e n e r g y ' . B e this as it
carbon
is
m a y , t h e r a t i o o f t h e total c a r b o n f l u x e d to p r e c u r s o r s
e x c r e t e d as a c e t a t e . T h i s g e n e r a t e s A T P a n d p r e v e n t s
u s e d f o r b i o s y n t h e s i s to t h e flux o f c a r b o n e v o l v e d
accumulation
T h e b u l k o f t h e c a r b o n d i o x i d e is m a d e b y o x i d a t i v e
as c a r b o n d i o x i d e ( P r e - C / C O 2) tells us s o m e t h i n g quite profound about the organisation of the CMPs.
d e c a r b o x y l a t i o n ( b u t P E P C c o n s u m e s C O 2) a n d t h e s e
T h o s e f e e d s t o c k s w h i c h f e e d into t h e p h o s p h o r y l a t e d
reactions
intermediates
dioxide and precursors and the balance
of low molecular weight compounds.
generate
reduced
nucleotides
which
are
of the
CMPs
have
a much
higher
Table 3 Inputs and outputs of the CMPs on eleven carbon sources Carbon source
Condition
Growth Uptake Acetate Carbon dioxide Precursor carbon ( m o l / k g / h ) Pre-C/CO z /* (h i ) ( m o l / k g / h ) ( m o l / k g / h ) ( m o l / k g / h )
Glucose 6-phosphate Batch 0.95 Glucose Batch/turbidostat 0.94 Gluconatc Batch 0.90 Glucumnate Batch 0,87 Fructose Batch 0,72 Glucose Chemostat 0,72 Glycerol Batch 0,70 Pyruvate Batch/turbidostat I).73 Fumarate Batch 0.63 Lactate Batch 0.60 Acetate Batch 0.43 Pyruvate Chemostat 0.30 Oxoglutarate Batch 0.32 Pyruvate Batch and bromopyruvate 0.30
11.76 10.57 1 1.48 11.74 7.62 6.22 14.1X) 42.87 16.52 23.66 20.16 12.50 6,65 12.98
4.(X1 4.89 6.48 9.07 0 0 11 16.81 0 4.06 - 211.16 0 0 1)
Pre-C/C02: flux to total precursor carbon/flux to net C02 evolution.
23.27 14.78 18.66 16.58 16.11 7.8t) 13.23 64.99 40.15 38.21 22.64 25.06 20.01 26.55
40.20 39.76 38.12 35.72 29.61 30.51 28.77 29.94 25.93 24.65 17.68 12.44 13.24 12.47
1.73 2.69 2.04 2.15 1.84 3.9 I 2.17 l).4~ !).65 0.65 0.78 I).511 0.66 0.47
102
14. Holms/FEMS Microbiology Reviews 19 (1996) 85-116
P r e - C / C O 2 ratio (1.7-2.69) than those which feed into the non-phosphorylated intermediates (0.550.78). This is true of those feedstocks which flux through the EDP and divide their carbon between phosphorylated and non-phosphorylated intermediates so that the destination of half the carbon confers this advantage of high P r e - C / C O 2. Compare glycerol (Fig. 10) and pyruvate (Fig. 12). The enzymes between PEP and fructose bisP are reversible so the interconversion of PEP and pyruvate is the only metabolic difference between the two. The direction of flux (PK vs. PEP synthetase) appears to have a very profound effect on P r e - C / C O 2 (2.17 vs. 0.55) On acetate, fumarate and OGA (Fig. 13, Fig. 21 and Fig. 22) flux to PEP and gluconeogenesis is by PEP carboxykinase which appears to be equally disadvantaged ( P r e - C / C O 2 of 0.78, 0.65 and 0.54). Overall the carbon sources entering the non-phosphorylated parts of the CMPs must overcome the problem of energy generation for gluconeogenesis. The gulf between the phosphorylated and the non-phosphorylated parts of the CMPs is more easily crossed in the direction towards the carboxylic acids which are the prime substrates for oxidative metabolism. One could perhaps anticipate this on the grounds that the phosphorylated pathways were themselves sufficient to support growth for the larger part of the evolution of the CMPs. It follows that when oxygen entered the ecosystem the task of redesigning the phosphorylated CMPs to take advantage of the hugely increased availability of ATP was too complex to be selected. The consequences of this will be discussed below (Section 3.3.) however it should be stressed that this argument applies to E. coli and there are other organisms (e.g. Pseudemonas and Acinetobacter species) which actually prefer carbon sources which enter the non-phosphorylated parts of the CMPs. It would be very interesting to measure P r e - C / C O 2 ratios for these organisms on different carbon sources. 3.2. Intervention to reduce acetate excretion
One very great advantage of presenting data as flux diagrams is that they illustrate where it might be possible to intervene and change metabolic fluxes. The flux diagrams for all those feedstocks which sustain acetate excretion (Fig. 7, Fig. 12, Fig. 16,
Fig. 18, Fig. 19 and Fig. 20) show certain features in common: the uptake of carbon source and flux into the CMPs exceeds their capacity to generate precursors and convert them to monomers; the flux through PDH produces more Ac.CoA than can be used for precursors or energy generation; phosphotransacetylase and acetokinase act as a safety valve to excrete intermediates from the CMPs; and the CMPs of all acetate excretion phenotypes are divided into two parts by the conversion of PEP to OAA. It is therefore possible that PEP carboxylase limits flux to OAA and OGA in these phenotypes. These possibilities will be dealt with in turn.
3.2.1. Excessive uptake
The easiest way to restrict the uptake of a carbon source is to restrict its availability and the simplest way to do that is in a chemostat. Glucose and pyruvate both excrete acetate but feed to opposite sides of the division in the CMPs. In both cases restriction of uptake in a chemostat (Table 3) can totally abolish acetate excretion (Figs. 14 and 15). In the glucose phenotype, acetate excretion falls to zero when the growth rate is 0.72 (76.6% of /Zm~x) but glucose uptake is reduced proportionally more (to 59%). The obvious effect on the CMPs is that PK is completely redundant (Fig. 14) and PDH falls to 41%. Glycolysis and the Krebs cycle are reduced by only slightly more than uptake. On pyruvate (Fig. 15), a much greater restriction must be imposed on uptake, down to 26%, to prevent acetate excretion. Growth is lowered to 42% as is flux through PEP synthetase (PEPsyn), PEPC and glycolysis all of which have only a biosynthetic function in this phenotype. PDH falls in line with uptake (to 26%) but flux round the Krebs cycle to 52%. In both cases reduction in uptake lowers growth rate and eliminates acetate excretion. The reduction required in pyruvate uptake is much greater than glucose because the enormous inward flux is also reflected in the largest acetate excretion seen on any carbon source which, on a molar basis, is no less than 47% of uptake.
H. Holms / FEMS Microbiology Reuiews 19 (1996) 85-116
3.2.2. Excessiue flux through PDH Clearly if the flux through PDH could be reduced this would diminish the amount of Ac.CoA available for excretion as acetate. 3-Bromopyruvate is an inhibitor of PDH with a well-known mode of action [15,16] which, added (50/xM) to a pyruvate batch culture, prevents flux to acetate excretion (Table 3). Growth rate (0.30), uptake (12.98 m o l / k g / h ) are rather more inhibited than in the chemostat but otherwise the effects on fluxes are similar [ 13]. One might be tempted to say that bromopyruvate inhibits uptake of pyruvate but it is probably more true to say that unrestrained PDH activity stimulates pyruvate uptake! 3.2.3. Cont;ersion of Ac.CoA to acetate It is commonly said that excretion of acetate is the reversal of its uptake but Jones et al. have shown [17] that, at least in K strains of E. coli, the enzymes of acetate uptake/excretion are constitutive (Table 4). It seems more sensible, therefore, to consider these primarily as a safety valve for acetate excretion. Be that as it may, it is relatively easy to raise mutants deficient in the acetate excretion enzymes by enriching those insensitive to fluoroacetate [ 1 8 20]. This mutant was derived from E. coli. ML308 [13] and grown on pyruvate. Flux analysis (Fig. 23) confirmed that acetate was not excreted but that lactate excretion was now employed to redress the balance of what was still an excessive pyruvate uptake. Clearly the acetate safety valve is very important and if it is closed, the biomass must find another one in order to survive when growing on a carbon source whose uptake cannot be controlled down to a level where the CMPs are able to cope with the fluxes generated.
103
Glucose6P ---"-~O~--~Monomenl Fructose6P
\/ TriosePs
~
~
Monomer,
I 3~rate
Pyruvate M....... ~
~
Monomers
Peno~yruvate ~
Monomers
L~
yruvat'e
AcOe~aal~te
_co,. y 1
I~
M.......
[
Citi'ate /
Lactate Ace~l.P~ ] A~P'~
isoCitrate
Acetate
GlO~Oat e
Excr!leExcl~e~e~ : o ~ Fig. 23. Flux analysis of a mutant of E. coli ML308, lacking transacetylase, on pyruvate.
The transacetylase negative mutant grows on pyruvate more slowly (/z = 0.63) than its parent (/x = 0.73). However for this fall in growth rate (to 80%) the uptake of pyruvate falls to 58% (Figs. 12
Table 4 Activity of enzymes for acetate uptake in E. coli PA239 and acetate excreted in E. coli ML308 Carbon source
Acetokinase ( m o l / k g / h ) PA239
Phosphotransacetylase ( m o l / k g / h ) PA239
Acetate excretion ( m o l / k g / h ) ML308
Glucose Glycerol Gluconate Ribose Pyruvate Lactate Acetate Malate
45 39 108 39 123 111 54 51
36 39 75 36 78 84 39 45
4.9 0 6.5 16.8 4.1 - 20.2 0
104
H. Holms / FEMS Microbiology Reuiews 19 (1996) 85-116
and 23). The deviation of pyruvate flux to excretion is 39% in ML308 but only 14% in the mutant. It is quite easy to see how the block in transacetylase shifts excretion from acetate to lactate but it is not so obvious how it diminishes pyruvate uptake. Nevertheless it certainly does so! The uptake of pyruvate could be affected by a change in the proton gradient or by a feedback loop from an intracellular pool of a metabolite closely related to pyruvate metabolism (pyruvate itself or Ac.CoA?). This conclusion illustrates both the strength and weakness of the technique of flux analysis. In this case it tells you that something has controlled pyruvate uptake but gives no indication of the mechanism.
3.2.4. Enhancement of flux to OAA and OGA The flux to OAA and OGA depends on the anaplerotic flux to the Krebs cycle by PEPC in all cases where acetate is excreted (Fig. 7, Fig. 12, Fig. 16, Fig. 18, Fig. 19 and Fig. 20). This enzyme has been cloned and then inserted into a strain of E. coli K12 [21]. This strain is very different from E. coli ML308 and grows much more slowly on all carbon sources (e.g. on glucose /z is 0.47 compared to 0.94 for E. coli ML308). The only fair comparison is with E. coli ML308 where, by the same technique, the available PEPC bas been increased, by a factor of 74.6 fold, in a constructed strain, E. coli ML308/JOE4 [22]. This strain was grown on glucose and the fluxes analysed (Fig. 24). The result was unexpected! The strain grew somewhat slower (and therefore had a slower flux through PEPC) took up less glucose, diminished acetate excretion by almost 60% and had zero flux through pyruvate kinase. Interpretation of this result must await further work but the principal effect of increasing expression of PEPC is to reduce uptake of glucose by the PTS and use the glucose uptake more efficiently. Presumably the genetic intervention activates some control of the glucose PTS by changing intracellular pools of intermediates or some similar mechanism. The result is half-way to that of the glucose limited chemostat (Fig. 15) with the most obvious similarity being the abolition of flux through PK. However, whatever may be the mechanism which restricts glucose uptake, it is not as efficient as the chemostat and acetate must be excreted.
= Glucose6P "~'-~--~-Monomer$
TS•
Glucose
F
6P
Fru~c@se bisp
Monomers
3PGlycerate ~
Monomers
PenolP[~yruvat~e
Monomers
17.60 1 o,ool 1 0.031
M....... ~ P y r u ~ a t e
Acetate ~
M. . . . . . .
t,OPAT~Acet ~ ateisoCitrao~o te ~ Excrete
Glutarl~_~ Monomers
Fig. 24. Flux analysis of a recombinant of enhanced in PEPC, growingon glucose.
E. coli
ML308,
3.3. Controls and limitations in the CMPs
The prime function of the CMPs is to produce precursors and the energy required to power their conversion to monomers and biomass. The PreC / C O 2 ratios fall into two divisions (Table 3) which suggests that this combined task is more easily performed if the starting point is one of the phosphorylated intermediates of the CMPs rather than a nonphosphorylated compound and reasons have been advanced which might explain this (Section 3.1.). However flux analysis can illustrate other fundamental features of the CMPs which can best be proposed as hypotheses which are listed here and then discussed in the sections shown: • acetate excretion balances excessive inputs with fluxes to precursors (Section 3.3.1.);
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
• when uptakes balance fluxes to precursors without any need for acetate excretion, uptakes are limited by feedback control or uptake capacity (Section 3.3.2.); the unique junction created during growth on acetate is regulated in a unique way (Section 3.3.3.); flux to energy supply does not limit growth on any of the carbon sources tested (Section 3.3.4.); maintenance energy is thermodynamic inefficiency (Section 3.3.5.); the importance of junctions is not necessarily dependent on pool sizes except in the case of isocitrate in the acetate phenotype (Section 3.3.6.); 'control' of the CMPs generally lies outwith the CMPs (Section 3.3.7.). 3.3.1. Excessive flux of inputs causes acetate excretion The evidence to support this contention has already been given. Consider a high and low PreC / C O 2 input: glucose and pyruvate. When inputs are restricted in a chemostat, acetate excretion falls and can be totally eliminated (compare Figs. 14 and 15 with Figs. 7 and 12). The greater the excessive flux in the unrestrained culture, the greater the restriction of input required to eliminate acetate excre-
105
tion. Furthermore elimination of acetate excretion on pyruvate, by bromopyruvate inhibition of PDH, was dependent on a major reduction in pyruvate uptake. When acetate excretion was made impossible by genetic intervention (Fig. 23), pyruvate uptake was reduced to 58% of the unrestricted flux but was still sufficient to require an excretion of lactate. Finally, while the amplification of PEPC has consequences which are difficult to understand, the fall in acetate excretion (from 4.89 to 2.00 m o l / k g / h ) is accompanied by a fall in glucose uptake (from 10.57 to 7.60 mol/kg/h). 3.3.1.1. Conclusion. Uptake of carbon source at a rate which exceeds the capacity of CMPS to flux to precursors, is balanced by excretion of acetate as the first choice or lactate if this option has been lost by mutation. 3.3.2. Balance between input and output from the CMPs depends on control of input Both fructose and glycerol feed into the phosphorylated part of the CMPs, have high Pre-C/CO 2 ratios and do not excrete acetate but have somewhat lower growth rates than other inputs of this type which do excrete acetate (Table 3). The uptake of glycerol (Fig. 25) is controlled by feedback inhibi-
Gluc~ose6P~-~2~-
M. . . . . . .
Glycerol
ATP
Fructose6P
_~
Glyce rokinas~ "~r-~-~
[~
Feedback inhibition
I
Fructose bisP
I 0.
.721
GlyceroP dehydrogenase
GlyceroP
~ FA
~, TriosePs ~
Monomers
H2
3PGlycerate ~
Monomers
yruvate ~
Monomers
Peno
Remaining m o n o m e r s and Krebs Cycle
Fig. 25. Control of growth on glycerol.
106
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
tion [23]. Entry of glycerol into the CMPs is by glycerokinase which is sensitive to non-competitive allosteric inhibition by fructose 1,6-bisphosphate. Regulation of glycerol metabolism also protects E. coli by limiting the intraceilular pool of dihydroxyacetone phosphate which otherwise can feed a nonphosphorylated pathway leading to the synthesis of toxic methylglyoxal [23]. However, for the production of methylglyoxal to reach lethal proportions, an insensitive glycerokinase must have its effect enhanced by derepression of enzyme synthesis and reduction of catabolite repression. Indeed when the only factor is insensitive glycerokinase, growth rate is 30% faster than in the wild-type [23]. It should be stressed that these experiments used K strains of E. coli whose growth rates on glycerol (0.3-0.4) are much slower than ML308 (0.7) perhaps because K strains were chosen, in the first instance, for ease of transfer of genetic material rather than for studies of metabolism. Nevertheless, what little work has been published on glycerol metabolism in ML308 [24] shows that it works on the same principle as K strains. Fructose supports slower growth than its peers (G6P, glucose, gluconate, etc.) in the high PreC / C O 2 group. There is no direct evidence that the slower growth is attributable to restriction of fructose uptake. However continuous cultures of E. coli ML308, operating as turbidostats on mixtures of glucose and fructose, do not use fructose if sufficient glucose is present to support growth. If less glucose is available, it is all used and sufficient fructose is metabolised concurrently to maintain the higher growth rate characteristic of glucose [25]. Even when glucose supplies only 30% of the carbon input and fructose 70%, the growth is at the higher rate characteristic of growth on glucose alone. Even when fructose supplies 92.5% of the carbon input (7.5% from glucose) the growth rate is still 91% of that sustained on glucose alone. The conclusion must be that fructose can sustain faster growth than that seen in a fructose-only batch or turbidostat and therefore some mechanism must restrict fructose entry to the CMPs under these circumstances. It is very unfortunate that acetate was not measured in this work and we do not know if faster growth on fructose causes acetate excretion! Finally mannitol, in E. coli ML308, sustains a growth rate (0.94) as fast as
glucose (B. Clark, personal communication) and its metabolism is virtually identical to fructose (3) which again suggests that input of fructose is restrained. There is no evidence available on possible limitation to fumarate utilisation but growth is relatively slow (0.63). This could reflect the intrinsic difficulty in reversing glycolysis and crossing the C 4 - C 3 gap on this input or a limitation in the uptake mechanism for fumarate. Oxoglutarate is not an inducer of the dicarboxylic acid permease but is a poor substrate and will only grow on dicarboxylic acid permease constitutive mutants (W.H. Holms, unpublished observations). This could explain the slow growth on OGA (0.32) if inadequate uptake restricted carbon delivery into the CMPs. 3.3.2.1. Conclusion. Restriction of uptake and delivery into the CMPs of some carbon sources (e.g. glycerol) restricts growth rate and obviates the necessity for any acetate excretion. This may also be true of other carbon sources (e.g, fructose and OGA) but others (e.g. fumarate)may have a maximal rate of uptake which does not reach the threshold required to trigger acetate excretion. 3.3.3. Regulation of growth on acetate Acetate is delivered into the Ac.CoA pool in the CMPs by acetokinase and a transacetylase. These enzymes are constitutive (Table 4) and act as a safety valve to prevent build up of intermediates when the CMPs are overloaded. The remaining enzymes for acetate metabolism, ICL, malate synthetase and ICDH-kinase/phosphatase, are repressed except when acetate is the sole source of carbon. ICL and malate synthetase convert 1 tool OAA and 2 mol Ac.CoA into 2 tool OAA. This metabolic sequence has been called the glyoxylate bypass because glyoxylate is an intermediate unique to its operation [3]. This is the only example among our eleven different carbon sources where a junction is created within the CMPs when ICL and ICDH compete for a common substrate. The first inkling (in 1971) that this was a competition was when the reversible inactivation of ICDH was lbund to be a prerequisite for growth on acetate [26]. This mechanism allows ICL to compete with ICDH so that the partition of flux at this junction meets the requirements of fluxing carbon
107
H. Holms / FEMS Microbiology Ret'iews 19 (1996) 85-116
Table 5 Flux through ICDH relative to available enzyme activity Carbon source ICDH flux ICDHactivity Activity/ (mol/kg/h) (mol/kg/h) tlux
from acetate to the precursors required for biosynthesis (Fig. 5). The principles on which this system works have been described for E. coli ML308 [14] but are applicable to other strains and indeed to other Gram-negative rods [3]. In essence, ICL has a rather low affinity for isocitrate and requires the concentration of isocitrate in the pool to be raised to about 0.6 mM [27] to sustain flux to glyoxylate and succinate (Fig. 13). ICDH has a much greater affinity for isocitrate and must be restrained. This is achieved by a reversible phosphorylation of ICDH by a bifunctional kinase/phosphatase which is co-expressed with the other enzymes required for growth on acetate. The reversible phosphorylation is the mechanism of reversible inactivation because the phosphorylated form of the enzyme has no activity. This mechanism is the primary determinant of partition of fluxes during growth on acetate (Fig. 26). Although flux through ICDH is restricted it still makes up 68% of the flux from isocitrate with the balance (32%) going through •CL. 5,7% of the primary input to Ac.CoA is used directly for biosynthesis. It takes 3 mol of Ac.CoA to make each OGA (6.7% of input) and two to make each of pyruvate (9.9%) and O A A (31.2%). The remainder (46.6%) is oxidised in the Krebs cycle but outputs to OGA, Pyr and PEP all involve decarboxylations and, when these are taken
Glucose 3.36 Fructose 5.65 Glycerol 4.68 Gluconate 4.12 Pyruvate 11.63 Oxoglutarate 2.36 Fumarate 8,12 Acetate 9,84 Acetate 1CDH 3,36 dephosphorylated
60,7 53,0 66.2 81,2 72,8 135.6 92.6 10.0 70.2
into account, 56.2% of the original input carbon is lost as carbon dioxide and 43.8% delivered into the precursor pools. The precision of the control exerted by ICDH phosphorylation is reflected in the ratio of flux to enzyme activity measured on eight of our carbon sources, Table 5 [14]. The ratio of ICDH activity in the biomass to flux through ICDH varies from six to 58 on seven carbon sources, which cover the complete range of P r e - C / C O 2, but on acetate alone is 1.02. This is the only phenotype in which flux through ICDH utilises the full available activity of the enzyme. This means that phosphorylation of
Monomers
"~"~Gluconeogenisis
M ....... " ~ Penol Pyruvate l Monomers~
Acetyl ,i CoA
Monomers ~ " " ~ ~ ~Pyruvate
Oxal - - ~
~ ~
~l'---Acetyl.P ~ ==Citrate
~'
Acetate
ADP ATP
Acetate
7 [ ~ iso
Citrate
oa,.
=LI I,--. /
JL
Succinate
18.1 9.4 14. I 12.4 6.3 57.5 11.4 1.02 20.9
~
Fig. 26. Control of growth on acetate.
,.=P)
,,, .,o
108
H. Holms/ FEMS Microbiology Reviews 19 (1996) 85-116
ICDH regulates its activity in the cell to balance the flux of isocitrate between ICL and ICDH. Although the flux through ICDH is restricted by phosphorylation, the measured flux is between 2 - 4 times greater than that on five other carbon sources and is in the same range as fumarate and pyruvate. The fact that fluxes are so much less than the available enzyme activity, in all phenotypes except growth on acetate, indicates that the fluxes are determined by the concentration of isocitrate available in the pool rather than the enzyme activity. It follows that pool sizes must be small and none can be detected except during growth on acetate. The ICDH k i n a s e / p h o s phatase dephosphorylates ICDH when its activity is not required (e.g. on adding its end-product) or when it cannot function (anaerobiosis or cyanide poisoning). The dephosphorylated activity measured under these conditions is 70.2 m o l / k g / h . This means that, to function, 86% o f the available I C D H must be inactivated by phosphorylation to permit growth on acetate. Put another way, if 86% of the I C D H was not phosphorylated flux through ICDH would be 70.2 m o l / k g / h and a large additional amount of acetate would need to be taken up and completely oxidised. 3.3.3.1. Conclusion. Growth on acetate is limited by partition o f flux at the junction where ICL and ICDH
compete for isocitrate. This junction is unique to growth on acetate. 3.3.4. Fluxes to energy conservation in the CMPs It is quite simple to sum all the fluxes for generation of [2H] and A T P in the CMPs from any flux diagram, divide the net figure by the appropriate /z and express the results as mol [2H] or ATP per kg biomass made (Table 6). If we now assume that: the same biosynthetic pathways are always used to convert a given precursor to its monomers, irrespective of the carbon source generating the precursor; and it takes the same amount of [2H] and A T P to synthesise each monomer and assemble it into biomass. THEN it follows that it always requires the same amount o f [2H] and ATP to make 1 kg biomass from the appropriate set of precursors. The problem is that the CMPs generate different A T P / [ 2 H ] relative to the carbon sources. It follows that, in some cases at least, some [2H] must generate A T P by oxidative phosphorylation. Unfortunately we do not know how much ATP is generated by each [2H] oxidised to water in each phenotype, but, as a minimum, we might assume that each [2H] could surely generate one ATP. The sum of [2H] and, A T P generated in the CMPs is then the minimum amount of 'energy'
Table 6 Fluxes to ATP, [2H], CO2 on various carbon sources and conditions Input
Condition
p~
[2H] (mol/kg)
ATP (mol/kg0
[2H] + ATP (mol/kg)
CO2 (mol/kg/h)
Pre-C/CO 2
Glucose 6-phospate Glucose Gluconate Glucuronate Fructose Glucose Glycerol Pyruvate Fumarate Lactate Acetate Pyruvate Oxoglutarate Pyruvate
Batch Turbidostat/batch Batch Batch Batch Chemostat Batch Turbidostat/batch Batch Batch
0.95 0.94 0.90 0.87 0.72 0.72 0.70 0.73 0.63 0.60 0.43 0.30 0.32 0.30
58.48 40.95 40.13 49.53 56.16 29.38 69.2 109.5 86.43 138.75 116.74 108.6 94.94 145.2
36.24 19.34 12.52 14.11 15.90 9.00 13.57 24.00 2.51 4.67 - 34.33 6.50 17.66 8.33
94.7 60.3 52.7 63.7 72.1 38.28 82.8 133.5 88.9 143.4 82.4 115. l 112.6 153.5
23.27 14.78 18.66 16.58 16.11 7.80 13.23 65.05 40.15 38.21 22.64 25.06 20.01 26.55
1.73 2.69 2.04 2.15 1.84 3.91 2.17 0.46 0.65 0.65 0.78 0.50 0.66 0.47
Chemostat Batch Batch/BrPYR
109
H. Holms / FEMS Microbiology Retqews 19 (1996) 85-116
available from the CMPs to generate 1 kg biomass from the precursors simultaneously generated by the CMPs. The precursors must be used in the amounts calculated, because we have measured, in biomass, the monomers made from them, but not necessarily all the [2HI and ATP need be consumed because of futile cycles, 'slip' or any other thermodynamic inefficiency which oxidises [2H] or hydrolyses A T P to generate heat. Indeed if the energetic input to make 1 kg of biomass is constant, it is obvious that growth on many carbon sources is less efficient than others. A m o n g the high P r e - C / C O 2 group every culture generates more [2H] + ATP to make 1 kg biomass than does glucose in a chemostat at /~ 0.72. This must mean that the glucose chemostat is the most thermodynamically efficient of all these cultures and releases less heat than the others. Of course, it does not follow that the glucose chemostat is the most efficient system possible, only that it is the most efficient of this set. In the low P r e - C / C O 2 group several phenotypes generate less ATP than the glucose chemostat but this is because these feedstocks feed into the non-phosphorylated pathways which generate less A T P by substrate-level phosphorylation than the phosphorylated pathways. Indeed the only reason that pyruvate gives more A T P than the glucose chemostat is the A T P made available by acetate excretion. However the low P r e - C / C O 2 group, by their point of entry to the CMPs, generate very large amounts of [2H] and the net sum of [2H] plus ATP is 2 - 4 times that of the glucose chemostat. The only
conclusion can be that, of all the carbon sources and conditions tested, the glucose chemostat at /z0.72 is the most thermodynamically efficient, that is to say generates the least heat. There is yet another difference to be noted between the low and high P r e - C / C O 2 group. The glucose chemostat is more efficient than the turbidostat and growth restriction progressively raises the P r e - C / C O 2 ratio from 2.69 to 4.07 at the growth rate when acetate excretion is zero. As the input of glucose restricts /x from 0.94 to 0.72, excretion of acetate declines to zero, 23% less glucose is used to make lkg biomass, 31% less carbon dioxide is evolved, 53% less A T P and 28% less [2H] are generated in the CMPs and P r e - C / C O 2 rises from 2.69 to 3.91 (Table 7). Pyruvate gives totally different results. As /z is restricted from 0.73 to 0.30 when acetate excretion is eliminated, 30% less pyruvate is used to make 1 kg biomass, carbon dioxide evolution falls a little and then recovers, less ATP is generated but [2H] is virtually unchanged. The PreC / C O 2 index is also stable and low. There is certainly a large difference in behaviour between glucose and pyruvate. Pyruvate feeds into the non-phosphorylated part of the CMPs which contain only carboxylic acids, the substrates for decarboxylation and oxidation. It is obviously a much simpler task for the CMPs to generate precursors and energy from carbon sources which feed into the phosphorylated parts of the CMPs. W e do not know if the other members of the high P r e - C / C O 2 group would be-
Table 7 Fluxes to acetate excretion, energy and precursors on glucose and pyruvate in chemostat cultures Carbon Source
/x
Uptake (mol/kg)
Acetate (mol/kg)
CO2 (mol/kg)
ATP (mol/kg)
[2H] (rnol/kg)
Pre-C/CO 2
Glucose Glucose Glucose Glucose Glucose Pyruvate Pyruvate Pyruvate Pyruvate Pyruvate Pyruvate Pyruvate
0.94 0.88 0.82 0.81 0.72 0.73 0.70 0.68 0.61 0.47 0.37 0.30
11.24 10.66 9.99 9.64 8.64 58.8 51.2 48.0 43.9 42.0 40.9 41.3
5.20 4.05 2.71 2.26 0 23.1 15.3 11.7 6.2 2.7 0.7 0.0
15.72 14.30 13.12 11.90 10.83 89.1 86.8 79.4 78.5 79.4 80.3 82.7
19.34 16.92 14.27 12.86 9.00 24.0 16.9 13.2 9.2 7.2 6.2 6.5
40.95 36.83 34.21 32.05 29.38 109.5 104.0 100.4 101.0 103.3 105.2 108.6
2.69 2.97 3.23 3.56 3.91 0.46 0.47 0.52 0.52 0.52 0.51 0.50
110
14. Holms/FEMS Microbiology Reviews 19 (1996) 85-116
have like glucose in a chemostat or the low PreC / C O z like pyruvate but it seems probable because the difference between glucose and pyruvate simply reflects a consequence of the structure of the CMPs and the fact that they were originally selected for anaerobic growth. The fundamental problem of all metabolism is that the same pathways perform both functions but, in the aerobic mode, any amount of reducing power can, in principle, be fed into the electron transport 'sink'. 3.3.4.1. Conclusion. The CMPs function to provide precursors and energy. On all carbon sources tested in batch or turbidostat culture they generate more ATP and [2H] than is required to convert the precursors, which they also generate, into new biomass. The CMPs work more easily on carbon sources which feed the phosphorylated parts of the CMPs. 3.3.5. Maintenance energy The mathematical analysis of chemostat data suggested that a fraction of catabolism did not contribute to growth and this is certainly also true of the growth of E. coli ML308 considered here. In 1965, Pirt [28] attributed the non-growth related energy utilisation to 'maintenance' of the existing bacterial structure and suggested it was independent of growth rate. This would be a very anthropocentric view of a biological phenomenon as a machine, were it not for the fact that our experience suggests that the harder (faster) machines are driven, the more maintenance they require. In 1976, Neijssel and Tempest [29] pointed out that the data were not incompatible with linear dependence of maintenance on /z which later evolved into their theory of 'slip'. There can be no doubt that E. coli ML308 devotes progressively less ATP and [2HI to non-growth purposes as its growth rate on glucose and, to a lesser extent pyruvate, is lowered to the point where acetate excretion is eliminated. Therefore, if we accept that there is such a thing as maintenance, we would support the proposition that it is directly related to growth rate. This would seem reasonable because maintenance is the replacement of bits of the machine that no longer function properly. As the biological machine is made up of polymerised monomers, maintenance presumably requires the detection and re-synthesis of defec-
tive polymers. If these can be degraded to their monomers and recycled the only input required is energy. However E. coli was selected in successive periods of 'feast and famine' [30]. This scenario demands survival of famine and control of feast. We already know that long-term survival of prokaryotes is enhanced by enforced idleness and we routinely impose that by dehydration or very low temperatures. The closer 'maintenance' is to zero at zero growth rate, the better the chance of survival. We must therefore accept variable 'maintenance'. Once you have done this there is no need to adduce maintenance at all. Maintenance energy is nothing other than heat-why not call it thermodynamic inefficiency? 3.3.6. Turnover of pools The pool of isocitrate in E. coli ML308 growing on acetate is 1.54 m m o l / k g [27] and the flux through this pool is the combined flux through ICDH and ICL which is 14.43 m o l / k g / h (Fig. 13). Therefore the pool of isocitrate, during growth on acetate, turns over 2.6 times per second. The pool of isocitrate is too low to measure in E. coli ML308 when growing on other carbon sources [27]. Pools measured in other strains of E. coli which grow much more slowly than ML308 may be misleading [31]. However one study [32] on E. coli VPI9 is relevant because it grew rapidly on acetate, glycerol, gluconate and glucose at growth rates very similar to E. coli ML308 and reported pools of some phosphory-
Table 8 P o o l s o f C M P i n t e r m e d i a t e s in
E. coli ( m o l / k g d r y w t ) g r o w i n g
on glucose, gluconate, glycerol and acetate Intermediate
Strain
Glucose
Gluconate
Glycerol
Acetate
G6P
VP19
2.97
0.26
6.21
7.02
F6P
VPI9
2.97
0.26
2.16
0.04
FDP
VP19
23.76
(/.26
21.60
9.72
6PG
VPI9
0.65
3.19
0
0
3PG
VP19
2.03
2.40
2.97
0.86
PEP
K
0.21
-
-
-
PYR
K
0.90
-
-
-
Ac.CoA
K
0.93
-
-
-
Citrate
K
30.0
-
-
isoCitrate
ML308
< 0.05
< 0.05
< 0.05
1.54
Oxoglutarate
K
1.10
-
-
-
Malate
K
3.60
-
-
14. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
lated intermediates but not PEP. Pools of PEP and some non-phosphorylated intermediates were measured in a K strain of E. coli growing on glucose [33]. If we assume that these pools (Table 8) are similar to E. coil ML308 growing on the same carbon sources then we can get some idea of the turnover of pools on at least four carbon sources (Fig. 27, Fig. 28, Fig. 29 and Fig. 30). 6-Phosphogluconate is found in cells growing on gluconate or glucose by not acetate or glycerol [33]. On carbon sources which generate G6P by gluconeogenesis (e.g. acetate and glycerol) the flux diagrams assume that flux into the PPP to generate C 4 and C s sugars is by the anaerobic pathway (Fig. 11) from triose phosphates and no 6-phosphogluconate is detected in
~ Glucose6P--'--~-~- M. . . . . . . F ~ 6P~ A TGlycerol ~ ADP "~ ~. GlyceroP <~1~ \/
" TriosePs~
~
.......
M. . . . . . .
3PGlycerate ~
Monomers
Pyruva ~t e~
Monomers
Penol
Monomers
1~Glucose6P~ M
111
Pyru~ate
Oxal ~
Acetate
M....... "~--1~-~~ ~
Monomers
/
18~FFFrur~0~6i P: Acetate TriosePs ~
Monomers
4~ 3PGl~rate"~'~5~--~-M...... PenolPyruvate ~
~ ACcetl ~ " ~
M. . . . . . . ~
Acetate
GlOxo utarate [
~
Excrete
Fig. 28. Turnover of pools in E. coil ML308 growing on glycerol.
Monomers
~ / AcOtale
T |
Ox~o Glutarate
~
=.~Succinate
Monomers
Fig. 27. Turnover of pools in E. coli ME308 growing on acetate. Numbers in open circular arrows are pool turnovers (times/min).
these cultures. Where the flux diagram assumes that the PPP is fed by the oxidative pathway via 6-phosphogluconate (e.g. gluconate and glucose) this intermediate is present in the biomass (Table 8). On all carbon sources, except acetate, the isocitrate pool is undetectable and the figures given are minimum rates which could be very much bigger. When acetate is the carbon source, flux (Fig. 27) through the isocitrate pool is much faster (14.43 m o l / k g / h ) than on the other carbon sources (3.364.68 m o l / k g / h ) but the turnover of the pool is much slower ( 1 5 6 / m i n ) than on the others ( > l 1 2 2 / m i n to > 1560/min). This results from the need for a high substrate concentration to make ICL sustain the necessary flux to glyoxylate and also requires the inactivation by phosphorylation of 94.5% of the ICDH (Table 5) to restrain oxidative decar-
112
H. Holms / FEMS Microbiology Reviews 19 (1996) 85-116
boxylation of isocitrate. The flux from glucose to PEP (Fig. 30) is successfully partitioned into four fluxes of which two are biosynthetic (3.2 and 16.1%) one catabolic (14.6%) and transport (66.0%). This very important junction is served by a very small pool which must be completely used and regenerated every 47 ms! The importance of any junction is not reflected in the size of the pool. Apart from fructose 6-phosphate in the acetate phenotype, none of the known pools used in gluconeogenesis (Figs. 28 and 29) is very small but there is no information for PEP available. Finally, the data on pool sizes are old and the analysis of the CMPs (both by flux analysis and others) would benefit from a complete survey of all
~_~FructosebisP TriosePs
~
127~Penot
Monomers
Pyruvate ~
Monomers
\PEP°
M....... ~ - - ~ 8 " ] ~ 3 ~
Ac~e/.~a~e "~-~
ate
Glucose6P~-[-'~-~,'- M. . . . . . . nomers
ate 10~:;TG ~ I~cOnate /
:
Monomers
+
Monomers
2Keto3deoxy /.4~.__~ gluconate 6P / / I 8,65I ~ 4 ~ 3'PGlycerate"mr'~'~-~ M. . . . . . . venoI
yruvate ~
iso Citrate
>11~
.tce~fe GlutaratOXTo e3~__
\/
\
Citrate~
AOp~ "
~ 1 ~ TriosePs ~
M. . . . . . .
.......
Fig. 30. Turnoverof pools in E. coli ML308growingon glucose.
Monomers
the intermediates of the CMPs growing on a number of carbon sources ! Monomers
Monomers
3.3.7. Control o f the CMPs ......
+ Excrete
Monomers
Fig. 29. Turnover of pools in E. coli ML308 growing on gluconate.
The function of the CMPs is to provide the precursors and energy for growth. The amount of each precursor is dictated by the monomeric composition of the biomass which, in turn, is dictated by the fraction of the DNA expressed in any phenotype. Although the composition of the cell varies with growth rate, this is most obvious when very fast growth rates are obtained in nutritionally rich media in which E. coli does not have to make many monomers de novo. We assume that when E. coli ML308 is grown on any of the eleven carbon sources, and it makes all its monomers de novo from that
14. Holms / FEMS Microbiology Ret,iews 19 (1996) 85-116
source, that the provision of precursors does not require to be varied over the relatively small range of /x involved. Further, we assume that the biosynthetic pathway from each precursor to each monomer is invariable and the amount of energy required to drive these, routes and then assemble the monomers into biomass is also constant. The biosynthesis of monomers and their polymerisation is very tightly controlled because we have never found any quantities of surplus materials accumulating in the cultures. It follows that the flux to precursors is exactly matched to the biosynthetic requirements. Lastly we conclude, from some evidence (Section 3.3.4.), that energy supply does not limit the processing of precursors into biomass and that a fraction of the potential 'energy' generated by flux through the CMPs is dissipated as heat. Why has some simple control mechanism not been selected to improve thermodynamic efficiency? The basic pattern of the CMPs was selected to cope with anaerobic metabolism when the provision of energy was probably limiting. When an E. coli ancestor acquired the ability to use [2H] to reduce molecular oxygen it increased its potential for ATP generation by an order of magnitude. However, to use all this energy for biomass generation it would have had to increase the flux to each precursor, in exactly the correct proportion, to take full advantage of the opportunity and this is the limitation. The provision of the correct mix of precursors is limited by the prehistory of the machine devised to make them which evolved for anaerobic metabolism. The relative fluxes from the precursors are fixed relative to each other by the spectrum of monomers required to generate biomass and this is dictated by the genome. It is tempting to say that the inability to sustain these fluxes aerobically without generating more 'energy' than is required to utilise them for growth is the fundamental reason for the thermodynamic inefficiency observed. However this is a hypothesis which can only be tested by further experiment. Because flux to all precursors would require to be changed simultaneously, the entire structure of the CMPs would require modification and this is clearly a difficult objective for selection. E. coli ML308 is forced to dispose of its excess [2H] by reducing something as it has always done throughout its evolutionary history. The conclusion
113
must be that, if there is any limitation within the CMPs, it must be to the supply of the complete range of all the precursors required for biosynthesis. In the simple overview of flux from carbon source to new biomass (Fig. i) the fluxes from CMPs to precursors must contain the flux which limits growth rate. Growth rate of each phenotype is probably limited by one unique flux to one precursor. Relief of this limitation would shift the limitation to the flux of another precursor, Jensen and Pedersen [34] considered the polymerase reactions which generate polymers and concluded that these were "subsaturated with precursors and catalytic components" and thus restricted growth rate. This conclusion is consistent with the view expressed here and explains how polymerisation reactions are controlled to cope with limitation in supply of precursors from the CMPs. The CMPs are not controlled to respond to the maximum potential demands of biosynthesis of monomers or the polymerase reactions for which they are the substrates. 3.3.8. Control theories Flux analysis describes the metabolic activity of the biomass, in particular the distribution of fluxes at metabolic junctions. It does not describe how these are controlled. There is no shortage of speculation to answer this question. Basically there are two theories which are usually presented in opposition to each other and have generated a vast literature which can best be summarised by two short reviews [35,36] published simultaneously. Kacser [35] has proposed a mathematical model of control theory in which the basic assumption is that there is a finite amount of control in any system and the sum of the control coefficients of all the enzymes in any system is defined as 1 (100%). When the effects of very small perturbations in steady states are examined experimentally, the conclusion is that 'control', defined in this way, is distributed throughout the system and it is rare to find any one enzyme reaction with a very high control coefficient. Even if a reaction with a control coefficient close to 1 is found it does not follow that this reaction is the 'pacemaker' for the whole system because the theory includes negative control coefficients. The significance of these cannot be overemphasised. If, for example, we are looking at control of flux to a product we might find that one
114
14. Holms / FEMS Microbiology Reciews 19 (1996) 85-116
enzyme on the route has a high control coefficient and we might be tempted to conclude that this is significant for control of flux to product. However if 'upstream' from this there is a junction at which flux is diverted to a product (excreted or not) that we have not yet discovered then that flux could well be the determinant, by a large negative control coefficient, of flux to product. The conclusion is obvious. If you do not know everything that the system is doing, you cannot start to work out how it is controlled. The solution is also obvious. You must analyse flux, that is measure everything that goes in and everything that comes out, before you can consider how the system is controlled. Newsholme [36] has called attention to the powerful allosteric controls exerted on irreversible enzymes in the CMP's. The supporters of Kacser's control theory have pointed out that, in the steady state, these enzymes do not seem to exert any degree of control much greater than other enzymes. It seems to me that both sides in the confrontation offer experimentally verifiable facts and it would be more sensible to reconcile them rather than express them in opposition to each other and I think this can be done very simply. In the steady state, fluxes through all pools and the sizes of these pools remain constant. For example, the excretion of acetate when E. coli is growing on glucose is a very good mechanism to achieve this. There is no need for any one enzyme in the whole system to be rigidly controlled and that is what is observed. In other words, the flux through glycolysis to precursors and to PEP is balanced with the subsequent flux to the non-phosphorylated precursors by excretion of acetate. Pyruvate dehydrogenase is the fulcrum of this balance. If acetate excretion has a large negative control coefficient it means that the glycolytic system itself does not require much control elsewhere. However a sudden perturbation to this steady state (such as shutting off supply of an input like carbon source or oxygen) could be catastrophic. If glycolysis continued (e.g. without oxidative generation of ATP), then the ATP pool would be depleted by generation of phosphorylated intermediates. When oxygen again became available, the biomass would not be able to resume normal activity because of unavailability of ATP and the biomass would be 'dead'. The intervention of allosteric control of irreversible enzymes (in this case,
fructose bisphosphate kinase) prevents this and this type of control has evolved to protect the pools of essential cofactors (ATP/ADP; NADPH,H+/NADP ÷) from sudden perturbations in the flow of metabolites. In the steady state there is no need for these emergency controls and the pool sizes of their effectors are such that these controls are not operative. Both Kacser and Newsholme are 'right' in the sense that their theories are valid explanations of metabolic control. These theories are not mutually exclusive but the mechanisms operate in different circumstances.
3.3.9. Deductions from .flux analysis E. coli ML308 grows over a 2.2-fold range of growth rates (0.43-0.95) on ten substrates or a three-fold range if you include oxoglutarate, the uptake of which is severely limited. Flux analysis suggests that energy supply does not limit growth rate and it follows that flux to precursors must be the limitation. When the carbon taken into the CMPs exceeds their ability to flux all of it to precursors, the excess is excreted as acetate. All of these phenotypes have ample energy supply but they must flux to precursors at different rates because they grow at different rates. G6P, glucose, gluconate, glucuronate, pyruvate and lactate all have enough carbon (because they excrete acetate) but they support a range of growth rates. It follows that the slower the growth, the slower the flux to precursors. All the precursors are required in defined amounts. If only one is limiting, the utilisation of all the others must he regulated down to the proportionate level to match the limitation. This must be one of the principle functions of the CMPs and those in E. coli ML308 certainly perform this task very well because we see no sign of oversupply to any precursor (except Ac.CoA). It is extremely probable that, in any one acetate excreting phenotype, flux to one precursor limits growth and flux to the others match this. If the flux to the limiting precursor could be increased, the flux to all the others would increase by the same proportion, at the expense of acetate excretion, and the growth rate would increase. An alternative experimental approach would be to supply the monomers derived from the limiting precursor which, if they were taken up and incorporated into polymers, would
H. Holms / FEMS Microbiology Reviews 19 (1996) 85 - 116
have the same effect-less flux to acetate and faster growth. However, the problem in sustaining flux to any one precursor must also reflect the metabolic problems to be overcome in fluxing carbon input to precursor. It must be 'easier' to make G6P from glucose than pyruvate and it follows that different phenotypes will have different limiting precursors. Experiments in which families of monomers derived from one precursor were fed to different phenotypes, in the steady state, would presumably test this hypothesis. Growth, utilisation of inputs and acetate excretion would all have to be measured. Satisfaction of one limitation in flux to precursor would probably create another. Presumably metabolic control analysis would produce similar results but it must be easier to add a family of monomers than manipulate enzyme amounts. In non-acetate excreting phenotypes the results of this type of supplementation would depend on whether (or how) the uptake of carbon source was regulated and the results would certainly illuminate these uncertainties. Lastly, the systems operating in E. coli ML308 were selected long ago under conditions vastly different from those that we impose upon it in the laboratory. It is generally accepted that 'feast and famine' [30] were the conditions in which the CMPs were selected. The three principal strategies which we observe today are control of uptake, excretion of surplus carbon and thermodynamic inefficiency. Control of uptake and excretion guard against the sudden onset of feast. The disadvantages they show in the laboratory are less than maximal achievable growth rate and profligacy with abundant carbon source. Presumably conditions of 'feast' were so rare that these were infrequent disadvantages and of such short duration that there was little pressure to select against them. The CMPs were originally selected to give precursors and energy under anaerobic conditions. Later, when oxygen became available, the system previously selected for anaerobiosis was suddenly gifted a new mechanism of oxidising reduced cofactors by a superabundant inorganic gas rather than an array of scarce organic compounds. Indeed the generation of oxygen also increased enormously the amount of organic material available. Selection under anaerobiosis was forced to balance carefully the metabolism of scarce resources among fluxes to
l 15
precursors and monomers, on the one hand, with those reactions trapping the energy required to convert monomers into biomass. Ability to access oxygen meant that one part of this balancing act (energy provision) was no longer a problem. So what we see today is a system to generate precursors which must dispose of excess reducing power by oxidation which generates heat. Growth on acetate is based on a uniquely different strategy in which competition between two enzymes is resolved by a unique mechanism. Presumably this is a later addition to the CMPs but it may be the vestige of an earlier glyoxylate oxidising cycle which was replaced by the Krebs cycle [37].
Acknowledgements I am grateful to the large number of colleagues who did the experimental work on E. coli ML308 and for their patience in discussion. However, this paper is dedicated to two friends who have died: Henry Kacser and Martin Smellie. Henry and I rarely agreed because we started with different attitudes but we always accepted each other's viewpoint and knew that our ideas would eventually be subsumed in a greater vision. In the meantime, we both enjoyed argument together. Martin supported and encouraged me, as he did for so many others. Without him I would have given up and certainly not extended the principles of flux analysis into the real world of fermentation biotechnology.
References [l] Mandelstam, J. and McQuillen, K. (Eds.) (1982) Biochemistry of Bacterial Growth (3rd Edn.) Blackwell Scientific Publications, Oxford. [2] Neidhardt, F.C., Ingraham, J.L. and Schaechter, M. (1990) Physiology of the Bacterial Cell. A Molecular Approach. Sinauer Associates, Inc., Sunderland, MA. [3] Neidhardt, F.C., lngraham, J.L. Low, K.B., Magasanik, B., Schzaechter, M. and Umbarger, H.E. (1987) Escherichia coil and Salmonella ~phimurium. Cellular and Molecular Biology, Vol. 1. Am. Soc. Microbiol., Washington, D.C. [4] Holms, W.H. (1986) The central metabolic pathways of Escherichia coli: Relationship between flux and control at a branch point, efficiency of conversion to biomass, and excretion of acetate. Curr. Top. Cellul. Regul. 28, 69-105.
116
H. H o l m s / F E M S Microbiology Reuiews 19 (1996) 85-116
[5] Dawes, E.A. and Senior, P.J. (1973) The role and regulation of energy reserve polymers in microorganisms. Adv. Microb. Physiol. 10, 135-266. [6] Dietzler, D.N., Leckie, M.P., Sternheim, W.L., Ungar, J.M., Crimmins, D.L. and Lewis, J.W. (1979) Regulation of glucose utilization in Escherichia coli during maintenance of the energy charge. J. Biol. Chem. 254, 8276-8287. [7] Neijssel, O.M. and Tempest, D.W. (1979) The physiology of metabolite overproduction. Symp. Soc. Gen. Microbiol. 29, 53-82. [8] Stouthamer, A.H. (1979) Microbial Biochemistry (Quayle, J.R., Ed.), pp. 1-47. University Park Press, Baltimore. [9] Roels, J.A. (1980) Application of macroscopic principle to microbial metabolism. Biotechnol. Bioengineer. 22, 24572514. [10] Tempest, D.W. and Neijssel, O.M. (1984) The status of ATP and maintenance energy as biologically interpretable phenomena. Ann. Rev. Microbiol. 38, 459-486. [11] Demain, A.L. (1972) Cellular and environmental factors affecting the synthesis and excretion of metabolites. J. Appl. Chem. Biotechnol. 22, 345-362. [12] Meyer, H.P., Leist, C. and Fiechter, A. (1984) Acetate formation in continuous culture of Escherichia coli K12 DI on defined and complex media. J. Biotechnol. 1,355-358. [13] E1-Mansi, E.M.T. and Holms, W.H. (1989) Control of carbon flux to acetate excretion during growth of Escherichia coli in batch and continuous cultures. J. Gen. Microbiol. 135, 2875-2883, [14] Holms, W.H. (1987) Control of flux through the citric acid cycle and the glyoxylate bypass in Escherichia coli. Biochem. Soc. Symp. 54, 17-31. [15] Bisswagner, H. (1981) Substrate specificity of the pyruvate dehydrogenase complex from Escherichia coli. J. Biol. Chem. 256, 815-822. [16] Lowe, P.N. and Perham, R.N. (1984) Bromopyruvate as an active-site-directed inhibitor of the pyruvate dehydrogenase multienzyme complex from Escherichia coli. Biochemsitry 23, 91-97. [17] Brown, T.D.K., Jones-Mortimer, M.C. and Kornberg, H.L. (1977) The enzymic interconversion of acetate and acetylcoenzyme A in Escherichia coli. J. Gen. Microbiol. 11)2, 327-336. [18] Brown, T.D.K., Pereira, C.R.S. and Stormer, F.C. (1972) Studies of the acetate kinase-phosphotransacetylase and the butanediol- forming systems in Aerobacter aerogenes. J. Bacteriol. 112, 1106-1111. [19] Guest, J.R. (1979) Anaerobic growth of Escherichia coli KI2 with fumarate as terminal electron acceptor: genetic studies with menaquinone and fluoroacetate-resistant mutants. J. Gen. Microbiol. 115, 259-271. [20] Levine, S.M. Ardeshir, F. and Ames, G.F.L. (1980) Isolation and characterization of acetate kinase and phosphotransacetylase mutants of Escherichia coli and Salmonella typhimurium. J. Bacteriol. 143, 1081-1085.
[21] Chao, Y.-P. and Liao, J.C. (1993) Alteration of growth yield by overexpression of phosphoenolpyruvate carboxylase and phosphoenolpyruvate carboxykinase in Escherichia coli. Appl. Environ. Microbiol. 59, 4261-4265. [22] Okungbowa, J. (1991) Genetic manipulation of metabolic fluxes in Escherichia coli. Ph.D Thesis, University of Glasgow. [23] Lin, E.C.C. (1976) Glycerol dissimilation and its regulation in bacteria. Ann. Rev. Microbiol. 30, 535-578. [24] Edgar, W., Forrest, I.S., Holms, W.H. and Jasani, B. (1972) The control of glycerol utilization by glucose metabolism. Biochem. J. 127, 59. [25] Clark, B. and Holms, W.H. (1976) Control of the sequential utilization of glucose and fructose of Escherichia coli. J. Gen. Microbiol. 95, 191-210. [26] Holms, W.H. and Bennett, P.M. (1971) Regulation of isocitrate dehydrogenase activity in Escherichia coli on adaptation to acetate. J. Gen. Microbiol. 65, 57-68. [27] EI-Mansi, E.M.T., Nimmo, H.G. and Holms, W.H. (1985) The role of isocitrate in control of the phosphorylation of isocitrate dehydrogenase in Escherichia coli ML308. FEBS Lett. 183, 251-255. [28] Pirt, S.J. (1965) The maintenance energy of bacteria in growing cultures. Proc. Roy. Soc. B. 163, 224-231. [29] Neijssel, O.M. and Tempest, D.W. (1976) Bioenergetic aspects of aerobic growth of Klebsiella aerogenes NCTC 418 in carbon-limited and carbon-sufficient chemostat culture. Arch. Microbiol. 107, 215-221. [30] Koch, A.L. (1971) The adaptive responses of Escherichia coli to a feast and famine existence. Adv. Microb. Physiol. 6, 147-217. [31] Holms, W.H., Hamilton, I.D. and Robertson, A.G. (1972) The rate of turnover of the adenosine triphosphate pool of Escherichia coli growing aerobically in simple defined media. Arch. Mikrobiol. 83, 95-109. [32] Moses, V. and Sharp, P.B. (1972) Intermediary metabolite levels in Eseherichia coli. J. Gen. Microbiol. 71, 181-190. [33] Lowry, O.H., Carter, J., Ward, J.B. and Glaser, L. (1971) The effect of carbon and nitrogen sources on the level of metabolic intermediates in Escherichia coli. J. Biol. Chem. 246, 6511-6521. [34] Jensen, K.F. and Pedersen, S. (1990) Metabolic growth rate control in Escherichia coli may be a consequence of subsaturation of the macromolecular biosynthetic apparatus with substrates and catalytic components. Microbiol. Rev. 54, 89-100. [35] Kacser, H. and Porteus, J.W. (1987) Control of metabolism: what do we have to measure? TIBS 12, 5. [36] Crabtree, B and Newsholme, E.A. (1987) A systematic approach to describing and analysing metabolic control systems. TIBS 12, 4. [37] Holms, W.H., (1986) Evolution of the glyoxylate bypass in Escherichia coli-an hypothesis which suggests an alternative to the Krebs cycle. FEMS Microbiol. Lett. 34, 123-127.